DEV Community: Brooke Jamieson

When Did Every AWS Service Launch?

Brooke Jamieson — Fri, 20 Mar 2026 20:49:31 +0000

Last updated: 20 March 2026

Last week I made a 6-minute video about the history of AWS and S3 to celebrate S3's 20th birthday (our baby is all grown up!). I did extensive research for this and as part of this I kept finding myself looking up things like "when did Lambda launch?" and "was DynamoDB before or after Redshift?" and getting frustrated that there wasn't a single page with all the dates in one place.

I've always wanted to put one together, but I was too lazy to manually go through hundreds of launch announcements across 20 years of What's New posts. But now we live in the future!

Then I started using the AWS MCP Server - a remote Model Context Protocol server that gives AI assistants authenticated access to AWS documentation, APIs, and pre-built workflows. One of its tools (search_documentation) lets you search across all AWS documentation, including the "What's New" announcements. This is super handy for keeping track of launches too - if you want to learn more, I have a whole blog on this called "You Should Just Use the Docs MCP to Keep Up with Launches During re:Invent Because Everything Else Is Too Annoying". (You can tell I did not use AI to write that title because it breaks every rule there is!)

So buckle up, here we go - every AWS service launch date I could find, from SQS in 2004 to Route 53 Global Resolver in 2026. Some services have both a preview and a general availability (GA) date so I've included everything that was available. And every launch links back to the original announcement so you can read it yourself.

Looking for a specific service?

Here are the most-searched AWS service launch dates. Click through to the full entry below for links to the original announcements.

Amazon SQS - 3 November 2004 (preview), 11 July 2006 (GA)
Amazon S3 - 13 March 2006
Amazon EC2 - 24 August 2006 (preview), 23 October 2008 (GA)
Amazon RDS - 27 October 2009
Amazon DynamoDB - 18 January 2012
Amazon Redshift - 14 February 2013
AWS Lambda - 13 November 2014 (preview), 9 April 2015 (GA)
Amazon ECS - 13 November 2014 (preview), 9 April 2015 (GA)
Amazon API Gateway - 9 July 2015
Amazon Aurora - 12 November 2014 (preview), 27 July 2015 (GA)
Amazon EKS - 29 November 2017 (preview), 5 June 2018 (GA)
AWS Fargate - 29 November 2017
Amazon SageMaker - 29 November 2017
Amazon EventBridge - 11 July 2019
AWS CDK - 11 July 2019
Amazon Bedrock - 13 April 2023 (preview), 28 September 2023 (GA)
Amazon Aurora DSQL - 3 December 2024 (preview), 28 May 2025 (GA)
Amazon Bedrock AgentCore - 21 July 2025 (preview), 14 October 2025 (GA)

What did AWS launch in 2004?

This is my favourite trivia question to ask user groups - did you know that the first AWS Service was SQS? It all started with a queue!

Amazon Simple Queue Service (SQS) - 3 November 2004 (preview) · announcement
A managed message queue service. SQS was the very first AWS service, launching in preview more than a year before S3 existed. It went GA on 11 July 2006. (GA announcement)

What did AWS launch in 2006?

S3 and EC2. Two services, and suddenly you could store objects and run virtual servers without owning hardware.

Amazon S3 - 13 March 2006 · announcement
Object storage. The service that started it all (well, after SQS). S3 turned 20 in 2026.

Amazon EC2 - 24 August 2006 (preview) · announcement
Virtual servers in the cloud. Launched in beta in 2006, went GA on 23 October 2008. (GA announcement)

What did AWS launch in 2008?

EC2 went GA, and EBS gave instances persistent storage for the first time.

Amazon Elastic Block Store (EBS) - 20 August 2008 · announcement
Persistent block storage for EC2 instances. Before EBS, your data disappeared when your instance stopped.

Amazon CloudFront - 18 November 2008 (preview) · announcement
Content delivery network. Went GA on 9 November 2010 with custom origin support. (GA announcement)

What did AWS launch in 2009?

Monitoring, managed databases, and networking. You could now run an app on EC2, put it in a VPC, store data in RDS, and watch it all with CloudWatch.

Amazon EMR - 2 April 2009 · announcement
Managed Hadoop (and now Spark, Hive, Presto, and more). Originally called Elastic MapReduce - the name tells you what era of big data this came from.

Amazon CloudWatch - 17 May 2009 · announcement
Monitoring and observability. Launched alongside Auto Scaling and Elastic Load Balancing.

Amazon VPC - 26 August 2009 (preview) · announcement
Virtual private networking. Originally optional - now every AWS resource lives inside a VPC by default.

Amazon RDS - 27 October 2009 · announcement
Managed relational databases. Started with MySQL, now supports PostgreSQL, MariaDB, Oracle, SQL Server, and (later) Aurora.

What did AWS launch in 2010?

SNS, Route 53, and IAM. Messaging, DNS, and the service that controls who can do what.

Amazon SNS - 7 April 2010 · announcement
Pub/sub messaging and push notifications.

AWS Identity and Access Management (IAM) - 2 September 2010 (preview) · announcement
The service that controls who can do what in your AWS account. Went GA on 3 May 2011. (GA announcement)

Amazon Route 53 - 6 December 2010 · announcement
DNS and domain registration. Named after the port number for DNS (53).

What did AWS launch in 2011?

CloudFormation, Elastic Beanstalk, SES, Direct Connect, and ElastiCache. Deploying and managing apps got a lot easier.

Amazon Elastic Beanstalk - 19 January 2011 · announcement
Platform-as-a-service for deploying web apps. You give it your code, it handles the infrastructure.

Amazon SES - 25 January 2011 · announcement
Email sending service.

AWS CloudFormation - 25 February 2011 · announcement
Infrastructure as code using JSON/YAML templates. The original IaC service on AWS.

AWS Direct Connect - 3 August 2011 · announcement
Dedicated network connection from your data centre to AWS.

Amazon ElastiCache - 22 August 2011 · announcement
Managed Redis and Memcached.

What did AWS launch in 2012?

DynamoDB, Glacier, SWF, Trusted Advisor, and Storage Gateway.

Amazon DynamoDB - 18 January 2012 · announcement
Managed NoSQL database.

AWS Trusted Advisor - 30 January 2012 · announcement
Automated best-practice checks for your AWS account.

Amazon Simple Workflow Service (SWF) - 21 February 2012 · announcement
Workflow coordination service. Largely superseded by Step Functions these days.

Amazon S3 Glacier - 20 August 2012 · announcement
Cold archival storage. Originally a standalone service, now integrated as S3 storage classes.

AWS Storage Gateway - 29 October 2012 · announcement
Hybrid cloud storage that connects on-premises environments to AWS.

What did AWS launch in 2013?

Redshift, CloudTrail, Kinesis, WorkSpaces, and a few others. Analytics and streaming showed up alongside the first virtual desktops.

Amazon Elastic Transcoder - 28 January 2013 · announcement
Media transcoding service.

Amazon Redshift - 14 February 2013 · announcement
Managed data warehouse. Launched on Valentine's Day, which feels appropriate for a service data teams love.

AWS OpsWorks - 18 February 2013 · announcement
Configuration management using Chef and Puppet.

AWS CloudHSM - 26 March 2013 · announcement
Hardware security modules in the cloud.

AWS CloudTrail - 13 November 2013 · announcement
API call logging. Records every action taken in your AWS account.

Amazon WorkSpaces - 13 November 2013 (preview) · announcement
Virtual desktops. Went GA on 26 March 2014. (GA announcement)

Amazon Kinesis Data Streams - 14 November 2013 (preview) · announcement
Real-time data streaming. Went GA on 16 December 2013. (GA announcement)

What did AWS launch in 2014?

Lambda, Aurora, ECS, Cognito, KMS, Config, and CodeDeploy. A big November - Lambda and ECS both previewed on the 13th.

AWS Cost Explorer - 8 April 2014 · announcement
Visualise and analyse your AWS spending over time.

Amazon Cognito - 10 July 2014 · announcement
User authentication and identity management for apps.

Amazon WorkDocs - 10 July 2014 (preview) · announcement
Enterprise document storage and sharing. Originally launched as Amazon Zocalo.

AWS Directory Service - 21 October 2014 · announcement
Managed Microsoft Active Directory in the cloud.

AWS CodeDeploy - 12 November 2014 · announcement
Automated code deployments to EC2, Lambda, and on-premises servers.

AWS Config - 12 November 2014 · announcement
Tracks configuration changes across your AWS resources.

AWS Key Management Service (KMS) - 12 November 2014 · announcement
Managed encryption keys.

Amazon Aurora - 12 November 2014 (preview) · announcement
MySQL and PostgreSQL-compatible relational database built for the cloud. Went GA on 27 July 2015. (GA announcement)

AWS Lambda - 13 November 2014 (preview) · announcement
Run code without provisioning servers. Went GA on 9 April 2015. (GA announcement)

Amazon Elastic Container Service (ECS) - 13 November 2014 (preview) · announcement
Managed container orchestration. Went GA on 9 April 2015. (GA announcement)

What did AWS launch in 2015?

API Gateway, IoT Core, OpenSearch (then Elasticsearch), WAF, Snowball, CodeCommit, CodePipeline, and more. Eighteen services across the year.

Amazon WorkMail - 28 January 2015 (preview) · announcement
Managed business email and calendar. Went GA on 4 January 2016. (GA announcement)

Amazon Elastic File System (EFS) - 9 April 2015 (preview) · announcement
Managed NFS file storage for EC2.

AWS Budgets - 29 June 2015 · announcement
Set custom cost and usage budgets with alerts.

AWS CodeCommit - 9 July 2015 · announcement
Managed Git repositories.

AWS CodePipeline - 9 July 2015 · announcement
Continuous delivery pipeline service.

AWS Device Farm - 9 July 2015 · announcement
Test mobile apps on real devices in the cloud.

AWS Service Catalog - 9 July 2015 · announcement
Create and manage approved IT service catalogues.

Amazon API Gateway - 9 July 2015 · announcement
Create, publish, and manage REST and WebSocket APIs.

Amazon OpenSearch Service - 1 October 2015 · announcement
Managed search and analytics. Originally launched as Amazon Elasticsearch Service, rebranded in 2021.

AWS WAF - 6 October 2015 · announcement
Web application firewall.

AWS Snowball - 7 October 2015 · announcement
Physical data transport device. Ship your data to AWS on a rugged appliance.

Amazon Kinesis Data Firehose - 7 October 2015 · announcement
Load streaming data into data stores and analytics services.

Amazon Inspector - 7 October 2015 (preview) · announcement
Automated security assessment. Went GA on 19 April 2016. (GA announcement)

Amazon QuickSight - 7 October 2015 (preview) · announcement
Business intelligence and data visualisation. Went GA on 15 November 2016. (GA announcement)

AWS IoT Core - 8 October 2015 (preview) · announcement
Connect IoT devices to the cloud. Went GA on 18 December 2015. (GA announcement)

Amazon Elastic Container Registry (ECR) - 21 December 2015 · announcement
Managed Docker container registry.

What did AWS launch in 2016?

Certificate Manager and DMS shipped in Q1, Flink arrived mid-year, and then November brought Lightsail, Athena, Polly, Rekognition, Step Functions, and a dozen more. Over 25 services this year.

AWS Certificate Manager - 21 January 2016 · announcement
Free SSL/TLS certificates for AWS services.

Amazon GameLift - 9 February 2016 · announcement
Managed game server hosting.

Amazon Lumberyard - 9 February 2016 · announcement
Game engine. (Later evolved into Open 3D Engine.)

AWS Database Migration Service (DMS) - 15 March 2016 · announcement
Migrate databases to AWS with minimal downtime.

AWS Application Discovery Service - 19 April 2016 (preview) · announcement
Discover on-premises applications for migration planning. Went GA on 12 May 2016. (GA announcement)

Amazon Managed Service for Apache Flink - 11 August 2016 · announcement
Real-time stream processing. Originally launched as Amazon Kinesis Analytics.

AWS Server Migration Service - 24 October 2016 · announcement
Migrate on-premises workloads to AWS.

AWS Organizations - 29 November 2016 (preview) · announcement
Centrally manage multiple AWS accounts. Went GA on 27 February 2017. (GA announcement)

Amazon Athena - 30 November 2016 · announcement
Query data in S3 using SQL. No infrastructure to manage.

Amazon Lightsail - 30 November 2016 · announcement
Simple virtual private servers with fixed monthly pricing.

Amazon Polly - 30 November 2016 · announcement
Text-to-speech service.

Amazon Rekognition - 30 November 2016 · announcement
Image and video analysis using deep learning.

Amazon Lex - 30 November 2016 (preview) · announcement
Build conversational interfaces (the tech behind Alexa). Went GA on 19 April 2017. (GA announcement)

AWS IoT Greengrass - 30 November 2016 (preview) · announcement
Run AWS Lambda functions on edge devices. Went GA on 7 June 2017. (GA announcement)

AWS Snowball Edge - 30 November 2016 · announcement
Snowball with local compute capabilities.

AWS Snowmobile - 30 November 2016 · announcement
A literal shipping container truck for exabyte-scale data transfer. My favourite AWS Service of all time.

AWS CodeBuild - 1 December 2016 · announcement
Managed build service for compiling and testing code.

AWS Personal Health Dashboard - 1 December 2016 · announcement
Personalised view of AWS service health events affecting your resources.

AWS Shield - 1 December 2016 · announcement
DDoS protection.

AWS Step Functions - 1 December 2016 · announcement
Visual workflow orchestration for distributed applications.

Amazon AppStream 2.0 - 1 December 2016 · announcement
Stream desktop applications to any device.

Amazon Pinpoint - 1 December 2016 · announcement
Targeted push notifications and campaign management.

AWS Glue - 1 December 2016 (preview) · announcement
Serverless ETL (extract, transform, load) service. Went GA on 14 August 2017. (GA announcement)

AWS Artifact - 7 December 2016 · announcement
On-demand access to AWS compliance reports.

AWS Managed Services - 12 December 2016 · announcement
AWS operates your infrastructure for you.

What did AWS launch in 2017?

Connect and X-Ray shipped in the spring. The Elemental media suite arrived in November. Fargate, SageMaker, EKS, and GuardDuty followed days later. Over 30 services this year - the most in a single year until that point.

AWS Batch - 5 January 2017 · announcement
Run batch computing workloads.

Amazon Chime - 13 February 2017 · announcement
Video conferencing and online meetings.

Amazon Connect - 28 March 2017 · announcement
Cloud-based contact centre.

AWS CodeStar - 19 April 2017 · announcement
Unified interface for managing software development activities.

AWS X-Ray - 19 April 2017 · announcement
Distributed tracing for debugging production applications.

AWS Migration Hub - 14 August 2017 · announcement
Track application migrations across multiple AWS tools.

Amazon Macie - 14 August 2017 · announcement
Discover and protect sensitive data using machine learning.

VMware Cloud on AWS - 28 August 2017 · announcement
Run VMware workloads on AWS infrastructure.

AWS PrivateLink - 8 November 2017 · announcement
Private connectivity between VPCs and AWS services without traversing the public internet.

AWS Amplify - 21 November 2017 · announcement
Build full-stack web and mobile apps. Started as a JavaScript library, now a full development platform.

AWS Elemental MediaConvert - 26 November 2017 · announcement
File-based video transcoding.

AWS Elemental MediaLive - 26 November 2017 · announcement
Live video encoding.

AWS Elemental MediaPackage - 26 November 2017 · announcement
Video origination and packaging.

AWS Elemental MediaStore - 26 November 2017 · announcement
Storage optimised for media.

AWS Elemental MediaTailor - 26 November 2017 · announcement
Server-side ad insertion for video.

Amazon GuardDuty - 28 November 2017 · announcement
Threat detection for your AWS account. Analyses CloudTrail, VPC Flow Logs, and DNS logs.

Amazon MQ - 28 November 2017 · announcement
Managed message broker for Apache ActiveMQ and RabbitMQ.

AWS AppSync - 28 November 2017 (preview) · announcement
Managed GraphQL APIs with real-time data and offline support.

AWS Fargate - 29 November 2017 · announcement
Run containers without managing servers. Serverless compute for ECS and EKS.

AWS IoT Device Management - 29 November 2017 · announcement
Onboard, organise, and remotely manage IoT devices at scale.

AWS Systems Manager - 29 November 2017 · announcement
Operational hub for managing AWS resources. Patch management, parameter store, run commands, and more.

Amazon Comprehend - 29 November 2017 · announcement
Natural language processing - extract insights from text.

Amazon FreeRTOS - 29 November 2017 · announcement
Operating system for microcontrollers.

Amazon Kinesis Video Streams - 29 November 2017 · announcement
Stream video from connected devices to AWS for analytics and ML.

Amazon SageMaker - 29 November 2017 · announcement
Build, train, and deploy machine learning models.

AWS DeepLens - 29 November 2017 (preview) · announcement
Deep learning-enabled video camera for developers.

AWS IoT 1-Click - 29 November 2017 (preview) · announcement
Trigger Lambda functions from simple IoT devices.

AWS IoT Analytics - 29 November 2017 (preview) · announcement
Analytics for IoT data.

Amazon Transcribe - 29 November 2017 (preview) · announcement
Automatic speech-to-text.

Amazon Translate - 29 November 2017 (preview) · announcement
Neural machine translation.

Amazon Neptune - 29 November 2017 (preview) · announcement
Managed graph database. Went GA on 30 May 2018. (GA announcement)

Amazon Elastic Kubernetes Service (EKS) - 29 November 2017 (preview) · announcement
Managed Kubernetes. Went GA on 5 June 2018. (GA announcement)

AWS IoT Device Defender - 29 November 2017 (preview) · announcement
Security management for IoT devices. Went GA on 2 August 2018. (GA announcement)

AWS Serverless Application Repository - 30 November 2017 (preview) · announcement
Discover and deploy serverless applications. Went GA on 21 February 2018. (GA announcement)

AWS Cloud9 - 30 November 2017 · announcement
Cloud-based IDE.

AWS IAM Identity Center - 7 December 2017 · announcement
Centrally manage single sign-on access to multiple AWS accounts. Originally launched as AWS SSO.

What did AWS launch in 2018?

Secrets Manager, Private CA, and Firewall Manager shipped in April. Then November and December brought a wave of preview announcements — many with GA dates stretching into 2019 and even 2020. Transit Gateway and the FSx family also arrived.

AWS Firewall Manager - 4 April 2018 · announcement
Centrally manage firewall rules across accounts.

AWS Private CA - 4 April 2018 · announcement
Managed private certificate authority.

AWS Secrets Manager - 4 April 2018 · announcement
Store, rotate, and manage secrets like database credentials and API keys.

Amazon Corretto - 14 November 2018 (preview) · announcement
No-cost, production-ready distribution of OpenJDK. Went GA on 31 January 2019. (GA announcement)

AWS Resource Access Manager (RAM) - 21 November 2018 · announcement
Share AWS resources across accounts.

AWS Transfer Family - 25 November 2018 · announcement
Managed SFTP, FTPS, and FTP for S3 and EFS. Originally launched as AWS Transfer for SFTP.

AWS DataSync - 26 November 2018 · announcement
Automated data transfer between on-premises storage and AWS.

AWS Global Accelerator - 26 November 2018 · announcement
Improve availability and performance using the AWS global network.

AWS RoboMaker - 26 November 2018 · announcement
Develop, test, and deploy robotics applications.

AWS Transit Gateway - 26 November 2018 · announcement
Connect VPCs and on-premises networks through a central hub.

AWS IoT Events - 26 November 2018 (preview) · announcement
Detect and respond to events from IoT sensors.

AWS IoT Things Graph - 26 November 2018 (preview) · announcement
Visually connect devices and web services for IoT applications.

AWS IoT SiteWise - 26 November 2018 (preview) · announcement
Collect and analyse industrial equipment data. Went GA on 9 July 2020. (GA announcement)

AWS Elemental MediaConnect - 27 November 2018 · announcement
Reliable live video transport.

AWS Cloud Map - 28 November 2018 · announcement
Service discovery for cloud resources.

AWS Inferentia - 28 November 2018 · announcement
Custom ML inference chip designed by AWS.

AWS License Manager - 28 November 2018 · announcement
Manage software licences across AWS and on-premises.

Amazon Elastic Inference - 28 November 2018 · announcement
Attach GPU-powered inference acceleration to any EC2 or SageMaker instance.

Amazon FSx for Lustre - 28 November 2018 · announcement
High-performance file system for compute-intensive workloads.

Amazon FSx for Windows File Server - 28 November 2018 · announcement
Managed Windows-native file system.

AWS App Mesh - 28 November 2018 (preview) · announcement
Service mesh for microservices.

AWS DeepRacer - 28 November 2018 (preview) · announcement
Autonomous 1/18th scale race car for learning reinforcement learning.

Amazon Managed Blockchain - 28 November 2018 (preview) · announcement
Create and manage blockchain networks.

Amazon Managed Streaming for Apache Kafka (MSK) - 28 November 2018 (preview) · announcement
Managed Apache Kafka.

Amazon Textract - 28 November 2018 (preview) · announcement
Extract text and data from documents using ML.

Amazon Personalize - 28 November 2018 (preview) · announcement
ML-powered personalisation. Went GA on 10 June 2019. (GA announcement)

AWS Security Hub - 28 November 2018 (preview) · announcement
Centralised security findings. Went GA on 24 June 2019. (GA announcement)

Amazon Forecast - 28 November 2018 (preview) · announcement
Time-series forecasting using ML. Went GA on 22 August 2019. (GA announcement)

Amazon QLDB - 28 November 2018 (preview) · announcement
Quantum Ledger Database - immutable, cryptographically verifiable transaction log. Went GA on 10 September 2019. (GA announcement)

AWS Outposts - 28 November 2018 (preview) · announcement
Run AWS infrastructure on-premises. Went GA on 3 December 2019. (GA announcement)

Amazon Timestream - 28 November 2018 (preview) · announcement
Managed time-series database. Went GA on 30 September 2020. (GA announcement)

AWS Well-Architected Tool - 29 November 2018 · announcement
Review your workloads against AWS best practices.

What did AWS launch in 2019?

CDK and EventBridge went GA in July. Control Tower shipped in June. Lake Formation arrived in August. December brought another batch of previews including Braket, Detective, and Kendra.

Amazon DocumentDB - 8 January 2019 · announcement
MongoDB-compatible managed document database.

AWS Backup - 16 January 2019 · announcement
Centralised backup across AWS services.

Amazon WorkLink - 23 January 2019 · announcement
Secure mobile access to internal websites and web apps.

AWS Ground Station - 23 May 2019 · announcement
Communicate with satellites. Yes, actual satellites.

AWS Control Tower - 24 June 2019 · announcement
Set up and govern a multi-account AWS environment.

AWS Cloud Development Kit (CDK) - 11 July 2019 · announcement
Define cloud infrastructure using familiar programming languages instead of YAML/JSON.

Amazon EventBridge - 11 July 2019 · announcement
Serverless event bus. Connect applications using events from AWS services, SaaS apps, and your own code.

AWS Chatbot - 24 July 2019 · announcement
ChatOps for AWS - get notifications and run commands in Slack and Teams.

AWS Elemental Appliances & Software - 7 August 2019 · announcement
On-premises video processing hardware and software.

AWS Lake Formation - 9 August 2019 · announcement
Build, secure, and manage data lakes.

AWS Data Exchange - 13 November 2019 · announcement
Find, subscribe to, and use third-party data in the cloud.

AWS Launch Wizard - 13 November 2019 · announcement
Guided deployments for enterprise applications like SQL Server and SAP.

AWS DeepComposer - 1 December 2019 (preview) · announcement
ML-powered music composition. A musical keyboard for learning generative AI.

AWS IAM Access Analyzer - 2 December 2019 · announcement
Identify resources shared with external entities.

Amazon Braket - 2 December 2019 (preview) · announcement
Quantum computing service. Went GA on 13 August 2020. (GA announcement)

Amazon EC2 Image Builder - 2 December 2019 · announcement
Automate the creation and maintenance of AMIs.

AWS Compute Optimizer - 3 December 2019 · announcement
ML-powered recommendations for right-sizing compute resources.

Amazon Augmented AI (A2I) - 3 December 2019 (preview) · announcement
Human review workflows for ML predictions.

Amazon CodeGuru - 3 December 2019 (preview) · announcement
ML-powered code reviews and application performance recommendations.

Amazon Detective - 3 December 2019 (preview) · announcement
Investigate security findings. Went GA on 31 March 2020. (GA announcement)

Amazon Keyspaces (for Apache Cassandra) - 3 December 2019 (preview) · announcement
Managed Cassandra-compatible database. Went GA on 23 April 2020. (GA announcement)

Amazon Kendra - 3 December 2019 (preview) · announcement
ML-powered enterprise search. Went GA on 11 May 2020. (GA announcement)

Amazon Fraud Detector - 3 December 2019 (preview) · announcement
Detect online fraud using ML. Went GA on 28 July 2020. (GA announcement)

AWS Wavelength - 3 December 2019 (preview) · announcement
Ultra-low latency applications at 5G network edges. Went GA on 6 August 2020. (GA announcement)

What did AWS launch in 2020?

Fewer net-new services than previous years. Network Firewall, CloudShell, MWAA (managed Airflow), and a batch of ML services. A lot of 2019 previews also went GA.

Amazon AppFlow - 22 April 2020 · announcement
Transfer data between SaaS apps and AWS without code.

AWS CodeArtifact - 10 June 2020 · announcement
Managed artifact repository for software packages (npm, PyPI, Maven, etc.).

AWS Snowcone - 17 June 2020 · announcement
The smallest Snow family device. Portable, rugged edge computing and data transfer.

Amazon Honeycode - 24 June 2020 (preview) · announcement
Build mobile and web apps without code using a spreadsheet-like interface.

Amazon RDS Proxy - 30 June 2020 · announcement
Managed database proxy that makes applications more scalable and resilient.

AWS App2Container - 1 July 2020 · announcement
Containerise existing .NET and Java applications.

Amazon Interactive Video Service (IVS) - 15 July 2020 · announcement
Managed live streaming.

AWS Distro for OpenTelemetry - 21 October 2020 (preview) · announcement
AWS-supported distribution of the OpenTelemetry project for distributed tracing and metrics.

AWS Glue DataBrew - 11 November 2020 · announcement
Visual data preparation tool - clean and normalise data without writing code.

AWS Network Firewall - 19 November 2020 · announcement
Managed network firewall for VPCs.

Amazon MWAA - 24 November 2020 · announcement
Managed Apache Airflow for workflow orchestration.

Amazon Monitron - 1 December 2020 · announcement
End-to-end machine monitoring system using sensors and ML.

Amazon ECS Anywhere - 1 December 2020 (preview) · announcement
Run ECS tasks on your own infrastructure.

Amazon Lookout for Vision - 1 December 2020 (preview) · announcement
Spot product defects using computer vision. Went GA on 24 February 2021. (GA announcement)

Amazon Lookout for Equipment - 1 December 2020 (preview) · announcement
Detect abnormal equipment behaviour for predictive maintenance. Went GA on 8 April 2021. (GA announcement)

Amazon DevOps Guru - 1 December 2020 (preview) · announcement
ML-powered operations service that identifies application issues. Went GA on 4 May 2021. (GA announcement)

AWS Proton - 1 December 2020 (preview) · announcement
Automated infrastructure provisioning for container and serverless deployments. Went GA on 9 June 2021. (GA announcement)

AWS Panorama - 1 December 2020 (preview) · announcement
Add computer vision to existing cameras. Went GA on 20 October 2021. (GA announcement)

AWS Audit Manager - 8 December 2020 · announcement
Continuously audit your AWS usage for compliance.

Amazon Lookout for Metrics - 8 December 2020 (preview) · announcement
Detect anomalies in business metrics. Went GA on 25 March 2021. (GA announcement)

Amazon HealthLake - 8 December 2020 (preview) · announcement
Store, transform, and analyse health data. Went GA on 15 July 2021. (GA announcement)

AWS CloudShell - 15 December 2020 · announcement
Browser-based shell with the AWS CLI pre-installed.

Amazon Managed Grafana - 15 December 2020 (preview) · announcement
Managed Grafana for operational dashboards. Went GA on 31 August 2021. (GA announcement)

Amazon Managed Service for Prometheus - 15 December 2020 (preview) · announcement
Managed Prometheus for container monitoring. Went GA on 29 September 2021. (GA announcement)

Amazon Location Service - 16 December 2020 (preview) · announcement
Maps, geofencing, and location-based services. Went GA on 1 June 2021. (GA announcement)

What did AWS launch in 2021?

App Runner, MemoryDB, FSx for OpenZFS, Fault Injection Simulator, Resilience Hub, and Elastic Disaster Recovery. Fewer services than the 2017–2019 peak, but this year had a lot of smaller feature launches that are not listed here.

AWS Fault Injection Simulator - 16 March 2021 · announcement
Chaos engineering - run controlled experiments to test resilience.

Amazon Nimble Studio - 28 April 2021 · announcement
Cloud-based content creation for visual effects and animation studios.

Amazon FinSpace - 3 May 2021 · announcement
Data management and analytics for the financial services industry.

AWS App Runner - 18 May 2021 · announcement
Deploy containerised web apps and APIs from source code or a container image.

AWS Application Migration Service - 18 May 2021 · announcement
Lift-and-shift migration to AWS.

Amazon MemoryDB for Redis - 19 August 2021 · announcement
Redis-compatible, durable, in-memory database.

Amazon FSx for NetApp ONTAP - 2 September 2021 · announcement
Managed NetApp ONTAP file system on AWS.

Amazon MSK Connect - 16 September 2021 · announcement
Managed Kafka Connect connectors for streaming data in and out of Apache Kafka.

AWS Cloud Control API - 30 September 2021 · announcement
A uniform API for creating, reading, updating, and deleting AWS resources.

AWS Resilience Hub - 10 November 2021 · announcement
Define, validate, and track the resilience of your applications.

AWS Elastic Disaster Recovery - 17 November 2021 · announcement
Scalable, cost-effective disaster recovery to AWS.

Amazon FSx for OpenZFS - 30 November 2021 · announcement
Managed OpenZFS file system.

AWS IoT TwinMaker - 30 November 2021 (preview) · announcement
Build digital twins of real-world systems. Went GA on 21 April 2022. (GA announcement)

AWS Mainframe Modernization - 30 November 2021 (preview) · announcement
Migrate and modernise mainframe applications. Went GA on 8 June 2022. (GA announcement)

AWS Private 5G - 30 November 2021 (preview) · announcement
Deploy and manage your own private 5G network. Went GA on 11 August 2022. (GA announcement)

AWS IoT FleetWise - 30 November 2021 (preview) · announcement
Collect and transfer vehicle data to the cloud. Went GA on 27 September 2022. (GA announcement)

Amazon VPC IPAM - 1 December 2021 · announcement
IP address management at scale.

AWS Cloud WAN - 2 December 2021 (preview) · announcement
Build and manage a global wide area network. Went GA on 12 July 2022. (GA announcement)

What did AWS launch in 2022?

CodeWhisperer arrived in preview mid-year just before I started my job with AWS. Trainium went GA in October. The end of the year brought Verified Permissions, Supply Chain, Security Lake, and Application Composer.

AWS Billing Conductor - 17 March 2022 · announcement
Customise your billing data for showback and chargeback.

Amazon GameSparks - 23 March 2022 (preview) · announcement
Game backend service.

Amazon EMR Serverless - 1 June 2022 · announcement
Run Spark and Hive without managing clusters.

Amazon CodeWhisperer - 23 June 2022 (preview) · announcement
AI-powered code suggestions. Went GA on 13 April 2023. (GA announcement)

Amazon Redshift Serverless - 12 July 2022 · announcement
Run analytics without managing data warehouse infrastructure.

AWS Wickr - 26 July 2022 (preview) · announcement
End-to-end encrypted messaging.

AWS Trainium - 10 October 2022 · announcement
Custom ML training chip. EC2 Trn1 instances went GA.

Amazon Verified Permissions - 28 November 2022 (preview) · announcement
Fine-grained authorisation using Cedar policy language. Went GA on 13 June 2023. (GA announcement)

AWS SimSpace Weaver - 29 November 2022 · announcement
Run large-scale spatial simulations.

Amazon Omics - 29 November 2022 · announcement
Store, query, and analyse genomic and biological data.

AWS Supply Chain - 29 November 2022 (preview) · announcement
ML-powered supply chain management. Went GA on 4 April 2023. (GA announcement)

AWS Verified Access - 29 November 2022 (preview) · announcement
Zero-trust access to corporate applications without a VPN. Went GA on 28 April 2023. (GA announcement)

Amazon Security Lake - 29 November 2022 (preview) · announcement
Centralise security data in OCSF format. Went GA on 30 May 2023. (GA announcement)

AWS Application Composer - 1 December 2022 (preview) · announcement
Visually design serverless applications. Went GA on 7 March 2023. (GA announcement)

Amazon CodeCatalyst - 1 December 2022 (preview) · announcement
Unified software development service. Went GA on 20 April 2023. (GA announcement)

What did AWS launch in 2023?

Bedrock went GA in September. Amazon Q previewed in November. Clean Rooms, VPC Lattice, and Payment Cryptography also shipped.

AWS Telco Network Builder - 21 February 2023 · announcement
Deploy and manage telco networks on AWS.

AWS Clean Rooms - 21 March 2023 · announcement
Analyse combined datasets without sharing raw data.

Amazon VPC Lattice - 31 March 2023 · announcement
Service-to-service connectivity, security, and monitoring across VPCs and accounts.

Amazon Bedrock - 13 April 2023 (preview) · announcement
Access foundation models from AI21 Labs, Anthropic, Cohere, Meta, Stability AI, and Amazon through a single API. Went GA on 28 September 2023. (GA announcement)

AWS Payment Cryptography - 12 June 2023 · announcement
Managed cryptographic operations for payment processing.

AWS AppFabric - 27 June 2023 · announcement
Connect SaaS applications for security and management.

AWS HealthImaging - 25 July 2023 · announcement
Store, access, and analyse medical imaging data at scale.

AWS HealthScribe - 26 July 2023 (preview) · announcement
Generate clinical documentation from patient-clinician conversations. Went GA on 27 November 2023. (GA announcement)

AWS Entity Resolution - 26 July 2023 · announcement
Match and link related records across data sources.

Amazon DataZone - 4 October 2023 · announcement
Data management service for cataloguing, discovering, and governing data.

AWS B2B Data Interchange - 27 November 2023 · announcement
Automate EDI (electronic data interchange) document transformations.

Amazon Q - 28 November 2023 (preview) · announcement
Generative AI assistant for software development and AWS operations. Went GA on 30 April 2024. (GA announcement)

What did AWS launch in 2024?

Eight new services including Deadline Cloud, App Studio, Nova, Aurora DSQL (in preview), and a lot of ‘feature launches’ that are not full services so they’re not on this list.

AWS Deadline Cloud - 2 April 2024 · announcement
Managed render farm for visual effects and animation.

AWS App Studio - 10 July 2024 (preview) · announcement
Build enterprise applications using natural language. Went GA on 18 November 2024. (GA announcement)

AWS End User Messaging - 29 July 2024 · announcement
Send and manage messages across SMS, MMS, and push channels.

AWS Data Transfer Terminal - 1 December 2024 · announcement
Physical locations for high-speed data uploads to AWS.

AWS Security Incident Response - 1 December 2024 · announcement
Triage and respond to security incidents in your AWS account.

Amazon Nova - 3 December 2024 · announcement
Amazon's own family of foundation models, available through Bedrock. Includes text, multimodal, and image generation models.

Amazon Aurora DSQL - 3 December 2024 (preview) · announcement
Serverless, distributed SQL database with active-active high availability. PostgreSQL-compatible. Went GA on 28 May 2025. (GA announcement)

Amazon S3 Tables - 3 December 2024 (preview) · announcement
Cloud object store with built-in Apache Iceberg support, optimised for analytics workloads.

What did AWS launch in 2025?

AWS Transform went GA in May for mainframe, VMware, and .NET workloads. Bedrock AgentCore previewed in July and went GA in October. December brought Nova Act, S3 Vectors, and the next generation of Nova models.

AWS Transform - 28 May 2025 · announcement
Automates modernisation of mainframe, VMware, and .NET workloads using AI. Launched with support for mainframe, VMware, and .NET workloads, with a custom transformation capability added at re:Invent 2025. (VMware GA) (.NET GA) (Custom GA)

Amazon Bedrock AgentCore - 21 July 2025 (preview) · announcement
Infrastructure for deploying and running AI agents. Includes runtime, memory, gateway, identity, and observability components. Went GA on 14 October 2025. (GA announcement)

Amazon Nova Act - 2 December 2025 · announcement
Build AI agents that automate browser-based UI workflows - form filling, search and extract, QA testing.

Amazon S3 Vectors - 2 December 2025 · announcement
Vector storage and querying built into S3.

What has AWS launched in 2026 so far?

We're only a few months in, but new services are already arriving.

AWS Elemental Inference - 24 February 2026 · announcement
AI-powered live video processing. Automatically generates vertical content and highlight clips for mobile and social platforms in real time, alongside encoding.

Amazon Connect Health - 5 March 2026 · announcement
Agentic AI for healthcare, built on Amazon Connect. Five AI agents handle patient verification, appointment management, patient insights, ambient documentation, and medical coding.

Amazon Route 53 Global Resolver - 9 March 2026 · announcement
Anycast DNS resolver across 30 AWS Regions. Resolves public and private domains from anywhere, with DNS query filtering and centralised logging. Previewed at re:Invent 2025.

What’s missing from this list?

I am definitely sure I have missed something on this list, but I hope you’ll let me know in the comments! There were so many to get through, but then there’s also the thing where AWS has a habit of launching features that blur the line between “new service” and “major feature of an existing service.” So, I decided to try and include things that would have their own console entry, or their own pricing page, or their own “What’s New” launch announcement. There’s also some services I deliberately left out because they’re features rather than standalone services (So things like S3 Intelligent-Tiering, Lambda Layers, EC2 Auto Scaling). And there’s probably others I left out because it was a skill issue on my part. Live laugh love launches!

If you spot a gap, please let me know - I am extremely online and you can find me on LinkedIn or X.

If you want to do your own research like this, the AWS MCP Server is a great place to start! You can use it in Kiro, Claude Code, Cursor, or whatever AI Agent is your current favourite. It’s the reason this post exists - I’d been meaning to compile this list for years and never had the time until I had a tool that could help me find 250 announcements without opening 250 browser tabs.

About the Author:

Brooke Jamieson is a Senior Developer Advocate at AWS, focused on AI agents and developer tools. A former fashion model turned mathematician turned AI Engineer, Brooke moved from Australia to New York City for this role. They make byte-sized tech content about AI and AWS, and you can find them on LinkedIn, Instagram or X.

Brooke's AWS Valentine's Day Cards 2026

Brooke Jamieson — Mon, 09 Feb 2026 21:25:53 +0000

Are you looking for romantic, scalable and highly available AWS-themed Valentine's Day Cards to send to your tru luv? Well, you're in the right place.

I make these every year, and it's always hands-down my favourite post of the entire year. It also happens to be my 1 shot to be funny in AWS-related content all year, so buckle up!

Compute

Amazon EC2 instance types come in hundreds of configurations with names like c7gn.xlarge, r6i.metal, and x8aedz. They are all very useful but I dread ever having to say them out loud on stage.

EC2 Spot Instances give you up to 90% off On-Demand pricing, but AWS can take them back with just two minutes' warning. Like love, they can be fleeting.

In EC2 security groups , 0.0.0.0/0 means open to the entire internet, which is a rookie error. Closing it down to a single trusted source is the security equivalent of making it Facebook Official.

EC2 key pairs use SSH keys for secure instance access. This means you need the private key to connect, so you shouldn't be sharing this with anyone.

AWS Graviton5 was launched at re:Invent 2025 and is AWS's most powerful and energy-efficient processor.

AWS Trainium3 is custom silicon AWS designed from scratch for AI training. Other times, people use general-purpose chips, but this is much more efficient and it's what your crush deserves for their generative AI workloads.

AWS Trainium2 chips are purpose-built for ML training and cost a lot less than GPUs for the same work.

EC2 Auto Scaling adds and removes compute capacity based on demand so your app never runs out of room to grow, just like your love.

Serverless

Lambda cold starts happen when a function hasn't been called in a while and needs to get set up before it can get to work - like setting up a kitchen before it's time to cook. The first request can be slow, but then it warms up.

Lambda functions have a maximum execution time of 15 minutes and are billed per millisecond.

Lambda Durable Functions can pause and wait for up to a year, and you don't pay a thing while they're waiting.

Amazon API Gateway sits in front of your backend, so every single request has to pass through it for authentication, throttling, and routing before anything reaches your services.

AWS Step Functions orchestrates workflows as state machines. So, "state" is both a workflow stage and an emotional condition, and then "step" is the actual technical term for a transition here.

In Amazon ECS task definitions, the "essential" boolean flag means if that container fails, everything stops.

Amazon EKS is a fully managed Kubernetes service, so AWS runs the control plane and you don't have to.

Databases

Amazon DynamoDB gives you single-digit millisecond response times at any scale, so this really is true love.

Aurora Multi-AZ keeps a replica ready to take over as the primary in under 30 seconds if something goes wrong.

RDS Multi-AZ keeps a standby copy of your database in a different Availability Zone, always in sync, always ready and waiting.

Aurora DSQL is a serverless SQL database that patches, scales, and manages itself, so you can stop babysitting infrastructure and spend your time planning dates.

Neptune is a graph database so data is stored as nodes connected by edges, and edges are literally called "relationships."

ElastiCache keeps your most-accessed data in memory so you get it back in microseconds, without repeat trips to the database.

Valkey is open-source, so this one is not an AWS card, but Madelyn asked for this card last year and it was one of my favourites, so it's back for 2026! Valkey is an in-memory key-value store supported by ElastiCache and MemoryDB, and it's the most permissive open source alternative to Redis.

OpenSearch is a search engine built to find the best match across massive amounts of data, and the top result is always you.

Database Savings Plans give you discounted pricing in exchange for a commitment to consistent usage over a 1 or 3-year term.

Storage

Amazon S3 stores data with 99.999999999% durability. That's eleven nines, which means you're more likely to lose almost anything else before you lose data in S3.

S3 Vectors can handle up to 2 billion vectors per index and uses similarity search to find the "nearest neighbour", which in this case would be your closest match out of billions.

Amazon S3 Glacier is archival storage designed for data you don't really look at, but need to hold on to for years/decades.

AI and Machine Learning

Nova Omni takes text, images, video, and speech as input, so however you talk to it, it gets what you mean.

Amazon Nova Act builds autonomous browser agents that can take real actions from your natural language instructions.

Nova Forge lets you grab a base Nova model and then train it on your own data until it becomes your own custom frontier model.

Bedrock AgentCore is a platform for deploying and running AI agents at scale. It has built-in runtime, tool integration, identity, observability, and other helpful fetures, including persistent memory across conversations and policy controls that keep agents within boundaries.

Strands Agents is an open-source SDK where you invoke an agent by simply calling agent("your message") - the DX is slick!

Strands Agents lets you build production-ready AI agents in just a few lines of Python.

Networking and Content Delivery

CloudFront caches your content at 600+ edge locations around the world, so wherever your users are, it's already there waiting.

A VPC (Virtual Private Cloud) is your own isolated network inside AWS, so it's where your resources live. Every account comes with a default one per region, and deleting it is a classic "you don't know what you've got till it's gone" moment.

Security, Identity, and Compliance

Kind of controversial, but you're supposed to say AWS IAM like eye-am, even though I still spell out I-A-M. Regardless of how you say it, it controls who can do what in your AWS account.

In IAM policies, the wildcard * means "access to everything", and you're really not supposed to use it, but I can see why people would want to make an exception for their tru luv.

Amazon GuardDuty uses ML and threat intelligence to spot suspicious activity in your AWS environment, usually before you even know something's wrong.

AWS WAF blocks bad traffic before it reaches your app. Things like SQL injection, XSS, bots, etc.

AWS KMS manages the encryption keys behind almost everything protected in AWS. Things like S3 objects, EBS volumes, or databases. Without it, your secrets are out in the open.

Management and Governance

CloudWatch monitors your AWS resources around the clock and pings you when something's off. It's probably a red flag if this is happening in your actual relationship.

AWS CloudTrail logs every single API call in your AWS account. So it knows who did what, when, and from where. It remembers everything, and it never judges.

CloudFormation rolls back failed deployments to the last working state. So, when something breaks, it takes you back to when things were good.

Messaging

Amazon SNS topics can have many subscribers, but filter policies let you route specific messages to specific recipients, so even with a crowd wanting to listen, only one person gets your message.

Developer Favourites

AWS Certifications were really instrumental in growing my AWS knowledge, and my AWS Career alongside this. I wish this cert was real.

The AWS Well-Architected Framework defines six pillars of cloud best practices. Calling someone "Well-Architected" is the highest engineering compliment, but probably don't say this to someone who is super new to the cloud.

Amplify handles the backend complexity of full-stack apps, with absolutely lush developer experience. It gives you auth, APIs, storage, and hosting, so it looks easy even when it's doing a lot.

Terraform has a notoriously steep learning curve and breaking changes between major versions, but like true love, it can be worth it.

Rust also has a famously steep learning curve thanks to its borrow checker, but AWS uses it extensively and provides an official SDK.

Last but not least, Kiro is my favourite agentic IDE, and it's the thing I have open on my computer at work the most! Also, it launched during my literal actual wedding ceremony, so that has to be a sign.

Wrap Up

That's all 50! Now go send one to your favourite person, or your favourite service, I won't judge.

If you've got ideas for cards I missed, drop them in the comments - I'm always looking for more. Happy Valentine's Day!

Before Your Agent Books a Vacation, It Has to Learn to Scroll

Brooke Jamieson — Tue, 20 Jan 2026 20:16:51 +0000

When your AI agent fails, it can be a trap to think you just need to fix the prompt. It's reflexive, and a bit like when you're working with a new intern. If they don't understand your instructions the first time, it's normal to go back and try to figure out a better way to explain what you want them to do before putting them through a full on new training plan. Refining the way you are 'prompting' also gives you lots of quick wins early on, so EVERYONE involved feels more confident, but you can get to a point of diminishing returns after a while.

This is the moment where re-framing the task instructions stops being 'enough'. With prompting for AI, you've probably used a prompt generator by this point, added some examples (and counter examples), added some rules and constraints, and maybe even threatened firing the agent in the prompt if you're really feeling spicy. But even after all this, it's totally possible that the agent will still get stuck on a little calendar widget, then lose its context an sanity halfway through a booking flow. Once there's a bit of momentum at play, its so easy to just take the front-end dropdowns at face value when they look like they've been updated, even though the backend state hasn't been updated. This is when it's time for some practice drills.

The Gap Between Proof of Concept and Production

I was a mathematics major, so one of my little joys is reading the Amazon Science blog and keeping up with the research coming out of Amazo research teams. Today when I was absolutely not procrastinating from doing a reporting task, I was reading the blog and saw a post from Amazon's AGI Lab that caught my eye called "The unseen work of building reliable AI agents". In the post, the researchers describe "normcore agents" which are systems who excel at monotonous interactions that are very boring but vrey crucial for reliable software.

Reading this blog changed from being procrastination to 'research' when I got to this line:

"Before an AI can plan a vacation, it must learn to scroll. Literally. It must learn how to scroll … and click … and tab … and select a date that's hidden behind a pop-up … and recover when a form silently resets."

because this feels so relevant to what I'm seeing developers struggle with when they're working on building their own agentic systems.

Asking an agent to "book my summer vacation" is a simple request that leads to a wild workflow with hundreds of itty bitty steps. The agent needs to wrangle airline reservtation systems that were built when I was still in primary school, then it needs to deal with hotel systems which are consistently inconsistent, then there's payments (with currency conversions and regional taxes), and loyalty programs, and all sorts of compliance checks. A lot has go right for you to be able to book, and it needs to be right every time.

"She'll be right, mate" is one of my favourite Australian-isms, and I say it to myself every time I'm going from dev to prod. Every time, I think it's going to be easy pasy, and everything SHOULD be ok, but then life happens. It always feels like getting an acne breakout before a big event when things go wrong, even though you think you've done everything right. A big misconception (which the Amazon Science blog did a great job of unpacking for me) was that it's not always enough to fix the entire thing with a new or updated prompt.

The Gym Metaphor

Amazon's AGI Lab builds what they call "RL gyms" which are reinforcement learning environments where agents practice atomic behaviours:

"Just as an athlete builds core stability by repeating fundamental movements under controlled conditions, an agent develops reliability by practicing the smallest units of interaction in repeatable, instrumented scenarios."

This is the difference between bodybuilders with big 'show muscles' and people who are functionally fit. (I am in neither group, congrats if you're in one of them.) You can get puffed up with impressive muscles that look great in a demo on stage, but when it's time to actually help your friend move their fridge, it comes down to core strength and flexibility with movement that you can only really get by practicing, or walking the walk.

So a gym, in this cotext, is where you can 'isolate a skill, vary it, stress it, and measure it.' And this all results in what the Amazon Science researchers called an "agentic substrate" which is a foundation layer of basic skills and capabilities that agents can build on to use for domain-specific tasks.

The Three Workouts

The Amazon Science blog describes 3 workouts that show what agents actually need to practice, so they have the foundational skills they need.

Workout 1: Calendars

Calendars and booking systems are hard because they seem so simple in theory - we use them all the time. But once you get into recurring systems that do bookings every 3rd Monday of the month, and then that Monday happens to be a public holiday, things get really annoying really quickly. There's also time zone changes, daylight savings start/end dates (which always trip me up, because I didn't ever have daylight savings growing up!), and different holidays observed in different locations. Automating meeting booking for a team spread across time zones can turn into a nightmare super easily.

On top of all that, you're dealing with widgets that can go haywire under zoom or hide behind other UI layers. Elements re-render mid-click. And don't get me started on how differently calendar components behave across browsers.

So what does the agent really need to learn? It needs to recognize a widget's current state, and recover when it drifts, then commit the correct date exactly once, and remember to verify that the backend registered the change.

Workout 2: Dropdowns

"A dropdown menu might appear to have been updated before the backend has actually processed the change."

Once you start looking for this mismatch, you'll never unsee it. I've spotted it in enterprise apps, government systems, even slick consumer sites that should know better based on how lush their UI branding is. The agent can see a dropdown update and think "my work here is done" but the backend might still be processing, and the UI just lied to them.

Things can look fine on the surface - social media has taught us all this! - but the actual system state under the shiny UI tells us what's REALLY happening. Before you take the slick interface at face value, it's time to get a healthy amount of trust issues and dig in to check that the action was registered in the backend properly.

Workout 3: Async Endurance

Long workflows are brutal. When I'm doing a big spec task with Kiro, I get tired of watching the agent work and clicking 'next task' but then I always remember how much more tired I'd be if I was actually routing the requests or doing the work myself. Once you've got some async steps chained together - things like search, filter, validate, maybe refresh a few times - and then each one can have its own timing quirks. Text fields start to fight with autosuggest dropdowns that haven't finished loading. Sometimes the backend just... fails. And then the page looks loaded, but half the data is missing.

This is where agents "hit the wall." I know I would hit a wall too. They hit the context window limit - which for AI agents can happen when you have to look through a lot of large code files over an enormous repo, or researching lots of entries in a sprawling knowledge base. The agent just runs out of room to remember what it was doing in the first place, let alone figure out how to do that well.

The hard part is staying aligned with the true state of the system across dozens or hundreds of steps.

When Near Enough Isn't Good Enough

The Amazon Science blog mentions that some of their engineers come from autonomous vehicles. In that world, "almost right" is the same as "unsafe." You don't get points for being close, and the stakes are high.

"Agents don't just produce outputs; they take actions inside live systems. They touch databases, initiate transactions, and modify system states. And when the output of a model is a real change in the world, reliability becomes non-negotiable."

If your agent is booking flights or modifying customer records, "works most of the time" isn't good enough. You wouldn't accept that from a human employee. Why accept it from an agent?

What Does "Success" Actually Mean?

The research talks about "formal verifiers" - basically, specifications that define exactly what successful completion looks like. The button got clicked? Cool, but did the thing actually happen?

"A workflow like 'send an e-mail,' for example, isn't declared successful just because a button appears to have been clicked; it's declared successful because exactly one new e-mail record exists in the database, and no unrelated records have been created, modified, or deleted."

That's the bar, and agents have to clear it "not once but thousands of times, under shifting timing, network, and UI conditions."

What Can You Actually Build Today?

The Amazon Science blog describes internal research. In this case, it's RL gyms that Amazon's AGI Lab uses to train agents. So what can you use right now?

Amazon Bedrock AgentCore Browser solves the "I need my agent to browse the web but I don't want to become a browser infrastructure company" part. If you try to do this yourself, it will probably look like:

Week 1: "I'll just use Selenium!" (seems easy)
Week 2: Fighting ChromeDriver version mismatches
Week 4: Debugging memory leaks and zombie processes
Week 6: You're now managing browser pools, IP rotation, security hardening, and scaling infrastructure, and you have a job you never wanted

AgentCore Browser handles all the gross parts, so you can manage your own context window (in addition to the model's!). You are in charge of writing the agent logic, then AWS runs the browser for you. Session recording and replay let you debug exactly the kinds of calendar/widget failures the research describes.

Amazon Bedrock AgentCore Evaluations is the "formal verifiers" part of this, but for production. Remember how the research said success isn't "the UI looked right" but "the system state matches the specification"? You can go in and define what success actually looks like for your workflows, and then keep checking that your agent passes those tests. There are 13 built-in evaluators for things like tool selection accuracy and goal success rate, and you can build your own.

Amazon Bedrock AgentCore Memory helps with the "running out of room to remember" problem. Short-term memory is "what did we just talk about 3 messages ago?" and it keeps track within a single session so users don't have to repeat themselves. Long-term memory is "this user prefers morning meetings and hates Mondays" where it extracts insights across sessions, not just raw logs. This is how agents stay coherent across long workflows without hitting context window limits. I also dig into this concept in a recent blog called "Why AI Agents Need Context Graphs (And How to Build One with AWS)"

Amazon Bedrock AgentCore Observability shows you what REALLY happened. It gives you sessions, traces, and then spans let you see exactly what the agent attempted, Vs what the backend actually did. Instead of guessing based on what the UI showed, this is where you can see the real story.

Your Workout Plan

The takehome tip: your agent needs practice.

Prompts are instructions, but practice is repetition under varied conditions until the behaviour becomes reliable. The Amazon Science research shows us that agents need to satisfy verifiers "thousands of times" before they're production-ready.

The gym metaphor is a good memory device - so use it next time you need to explain this to the rest of your dev team. Your agents need isolated practice environments where failure is safe, varied conditions that stress-test edge cases, formal verification that defines what success actually means, and repetition until reliability becomes automatic.

This works for any agent framework, and also for humans. So, before your agent books a vacation, teach it to scroll.

The research referenced in this article comes from "The unseen work of building reliable AI agents" by Jason Laster, published on Amazon Science in January 2026. The AWS service descriptions are based on Amazon Bedrock AgentCore documentation.

Why AI Agents Need Context Graphs (And How to Build One with AWS)

Brooke Jamieson — Mon, 12 Jan 2026 17:33:09 +0000

Back in 2020, I wrote that dashboards tell you what's happening, but not why. That the rush to be "data-driven" was derailing more decisions than it was informing. Over five years later, we're building AI agents and they're hitting the same wall. We know that past decisions were made, but we have no idea why - context graphs are here to change that!

I finally had time to catch up on reading over the holidays (which felt SO LUXURIOUS!) - two posts really caught my eye: Jaya Gupta and Ashu Garg's piece on context graphs from Foundation Capital, and Animesh Koratana's technical deep-dive on how to actually build one. They gave a name to something I'd been circling for a while and now that I'm back at my desk, the first thing I wanted to do was write a practical guide for developers who want to build this with AWS.

Their posts are worth reading in full, and in this piece I wanted to dig in to the practical side: how do you actually build a context graph with the tools we have today?

What People Actually Want

Working in AI consulting taught me that a lot of the time when someone says "we need AI," what they actually mean is "we make the same kinds of decisions over and over, and we want to stop winging it." People would often think they wanted a magic wand or something, but they really just wanted some clarity in the rules and a nice way to handle the exceptions. Basically they want a decision tree with memory, but not just any memory. They need to remember why decisions were made, not just that they were made. And that reasoning needs to be accessible at the right moment, when a similar decision comes up again.

This is where the "data-driven" promise breaks down. Having data isn't enough. You need the right data, accessible at the right time. The same is true for memory. And it turns out, the kind of memory that matters most was never captured as data in the first place. Enterprise systems are gnarly and are full of old tools, weird workflows, SQL Server running on a server hidden under a staircase, random spreadsheets, Slack threads from 2021. For sure, some of it is straight up junk, but lots of this clutter is where the work really gets done. There's hidden gems where people store context, judgement and workarounds.

But the hidden gems are all HIDDEN! They haven't been captured or stored as data, so the reasoning connecting data to action has all fallen by the wayside. The AI agents we're building don't know about the whispered approvals or the "we tried this before and it blew up" pieces of the puzzle. None of this useful reasoning (which is the most important part of onboarding an employee) exists in a way that can be queried.

What Are Context Graphs?

A context graph captures decision traces. Not just WHAT happened, but WHY it was the right call, who approved it, what precedents informed it, and what alternatives were considered.

Think of it like git. Git captures what changed (the diff) and who changed it (the author - or git blame, if you're feeling scandalous). Git gives you a lot of info, but it misses a lot too - things that might have been discussed in standups or design discussions, the alternatives you considered but didn't end up going with, the tradeoffs you looked at, or why the 'winning approach' won.

Context graphs can help to capture the missing layer of reasoning.

Animesh frames this in a "two clocks" problem: therere's so much infrastructure for the state clock (what's true right now), but barely anything for the event clock (what happened, in what order, with what reasoning).

Why "Just Add Memory" Isn't Enough

If remembering things is so important, why not just add some memory? Well - memory alone isn't enough:

It's not just what you talked about, but also how you decided
The decisions need to be connected to the customrs, systems or services they touched
It's easy to think about searching conversations, but searching precedent is a whole different thing
Reasoning like "what would happen if..." is what was always missing from dashboards, and it's the same thing here.

Context graphs look at this in a different way. It's not memory like "what did we talk about," but memory as in "what did we decide, why, and what did it affect."

The "graph" part is what makes this more than a log. It's all about connections! So you could start with a customer and see everything that affected them, or start by looking at a problem and find similar things you've tackled before.

These connected traces become a record that you can query to look at HOW your org makes decisions. So it's not just what the policies are, but how they get applied in practice.

Building Context Graphs with AWS

Honest answer here is that I think Strands and AgentCore are the coolest/most useful things AWS has launched in a long time, but they don’t get talked about enough, and this is an example where they’re super useful! It's their time to shine!

An important caveat here is that context graphs are early. Not just on AWS, but everywhere! Jaya and Ashu's blog came out on Dec 22 2025, so you're early to the party. Animesh's post from Dec 28 2025 goes into structural embeddings, 'what if' simulation, and a self-discovering structure is also fresh! I wasn't the only person on PTO at the time, so I doubt anyone has productionized it yet. (If you have, please get in touch!)

So we're figuring this out together, but AWS has the building blocks you need:

What You Need	AWS Service	How It Helps
Agents that can call tools & figure out what to do next	Strands Agents SDK	Your agent decides which tools it should call, and what order to call them in
Remember what happened and why	AgentCore Memory	Stores facts, episodes, and summaries, not just the chat history
Talk to the APIs you already have	AgentCore Gateway	Uses MCP to connect agents to your systems
Know what's allowed, and log why	AgentCore Policy	Cedar policies with full audit trail telling you what was allowed/denied and why
Track who made each decision	AgentCore Identity	The "who" travels with the reequest (OAuth with identity propagation)
Debug and audit the whole chain	AgentCore Observability	OpenTelemetry traces linking identity → policy → tool → outcome

A note on costs before you build: Obviously you should use the AWS Pricing Calculator and the pricing page but tl;dr AgentCore Memory costs scale linearly. As a reference point, 100K short-term events + 10K long-term memories + 20K retrievals = $42.50/month.

I know I keep saying "it's not just chat history" and to drill down into this more, here's the strategies in play:

Semantic memory stores facts about things: "Service X has had 3 incidents this month"
Episodic memory tells you the full story of a decision: think of this like answering job interview questions in the STAR (Situation, Task, Action, Result) format.
Summary memory is what it sounds like: it's a condensed summary that's faster to recall and search when you need it.

Special shoutout to the episodic strategy, because it does something that maps directly to what Animesh described as "learning from trajectories." There's also a good reason that so many companies tell you to answer job interview questions with the STAR format - it works! In this case, each episode captures important context in a structured way, so you know what the agent was trying to do in the first place, but also what it did, and how everything turned out in the end.

This is also a good opportunity to go back and talk some more about things we learnt from Data Driven Decision Making because I think there's a good parallel here! One of the other things I learnt early in my data science career was that lots of companies just want to hoard data like a big dragon sitting atop a pile of gold - they don't use it, they just want to know they have it. This is often completely useless, because what's the point of having all this data if you're not going to use it for anything helpful? Reflections are what saves you from doing this all over again. As you accumulate episodes, AgentCore analyses them to find the INSIGHTS that businesses froth over, so you can see the patterns, best practices and lessons you (hopefully) learnt.

How Knowledge Compounds

I talked about hoarding data earlier, and knowledge is a key example of this because it gets SO MESSY over time if you keep shoving it in a garage and closing the door to hide it. Facts change over time, and people might describe the same fact in different ways anyway so this can all get very hectic very quickly.

This is why you need memory consolidation to tidy it all up. AgentCore Memory doesn't just keep appending things, it intelligently merges related memories, resolves any conflicts that pop up, and keep a trail of what's changed.

There's a really good AWS Labs repo with an SRE agent use case that shows this in action. When the Kubernetes agent figures out that a particular memory leak pattern is causing OOM kills, that knowledge is stored. Then, the next time any agent (not just the Kubernetes agent!) sees a similar issue, that pattern is available.

I think this gets at the "compounding" that Animesh was talking about - each investigation helps to make future investiations smarter.

Who Approved What?

Remember git blame? You can use AgentCore Policy and **Identity **together to make a more useful (and hopefully less passive aggressive) version of this that shows the full story of who requested it, what policy said it was OK, and why. Getting an audit trail that shows the connections between identity → policy evaluation → tool execution → outcome is super useful if someone asks "why did the agent restart that pod?", because you can trace it all back to: "Alice requested it, the SRE-actions policy permitted it because she has the on-call role, and here's what happened next."

It's important to note that reasoning and governance get split up here - reasoning lives in Memory and governance lives in Policy logs. I think this might be handy for compliance-heavy workflows/industries.

A Minimal Example

I've yapped a lot so far, so let's make things a bit more concrete. Here's a basic outline of what a decision-capturing agent with Strands might look like:

from strands import Agent
from strands.tools import tool
from strands.models import BedrockModel

# Define a tool that captures the decision trace
# This is where you'd store to AgentCore Memory in practice
@tool
def log_decision(
    situation: str,
    context_gathered: list,
    recommendation: str,
    reasoning: str
) -> str:
    """Log a decision trace for future reference.

    Args:
        situation: What triggered this decision
        context_gathered: Data points collected before deciding
        recommendation: The recommended action
        reasoning: Why this recommendation makes sense
    """
    # In practice: store in AgentCore Memory with a namespace
    # like /decisions/customers/{customerId} etc
    return f"Decision logged: {recommendation}"

# Choose your model via Bedrock
model = BedrockModel(model_id="anthropic.claude-sonnet-4-5-20250929-v1:0")

# The system prompt is the key here
# It tells the agent to capture reasoning, not just complete the task
agent = Agent(
    model=model,
    system_prompt="""You help make decisions and capture your reasoning.

    For every decision:
    1. Describe the situation
    2. List what context you gathered
    3. State your recommendation
    4. Explain your reasoning

    Always call log_decision before responding.""",
    tools=[log_decision]
)

The system prompt is what turns this from a task executor to a context graph builder.

What a Decision Trace Looks Like

Earlier in the piece, I talked about a policy example with Alice & Carol from the SRE Agent repo. When Alice (an SRE) and Carol (an exec) both look in to the same API degradation issue in different ways, they both get the same technical findings - it was because of a database config failure, response time increased by 33x, and memory exhaustion. But, the traces show different reasoning behind getting to the same insight:
Alice's trace:

Root Cause: Database configuration failure causing connection timeouts - 
missing ConfigMap 'database-config' and invalid permissions on data directory

Escalation: alice.manager@company.com or sre-oncall@company.com if resolution exceeds 1 hour
Notifications: #alice-alerts, #sre-team

Next Steps:
1. Immediate (< 1 hour): Create/restore missing 'database-config' ConfigMap 
   and fix permissions on database data directory
2. Short-term (< 24 hours): Increase Java heap space allocation and implement 
   connection pooling with proper timeout handling
3. Long-term (< 1 week): Optimize slow query and implement circuit breakers

Carol's trace:

Root Cause: Database service failure due to missing ConfigMap 'database-config' 
in production namespace, causing cascading failures

Escalation: Notify executive team if not resolved within 20 minutes
Notifications: Executive channels (critical severity only)

Next Steps:
1. Immediate (< 1 hour): Create/restore missing ConfigMap, fix permissions, 
   increase memory allocation
2. Short-term (< 24 hours): Implement circuit breakers to prevent cascading failures
3. Long-term (< 1 week): Review memory usage patterns in UserService.loadAllUsers

So they both noticed the same thing happening, and they both dug into it, and then the trace shows how it got the 'insight' including why it was handled this way, for this person, in this context.

Start Small

Remember - this is new - REALLY NEW - so we're all just figuring it out as we go. You do not need to do all of this on day one, and at this stage it's more than enough to just be aware that this is even a thing that we can almost do.

If you want to get stuck in, I recommend picking one workflow that's super decision heavy. Something where "it depends" is the honest answer. Start there, and then:

Make an agent to handle that workflow with Strands
Configure memory to capture those insights from the decision trace. To do this, you'll need to design namespaces around the things you care about (things like inputs, reasoning, outcome)
Build precedent lookups so you're not reinventing the wheel every time. You want to look at similar decisions from the past before making a new decision.
Iterate! This is where the 'figuring it out' part is key. Think about what you're missing, what would make this more helpful, and how you can continue to get value as you scale.

Where we're at now, and where we're going

Ok so - the "trillion-dollar opportunity" that Foundation Capital described in their blog isn't something I can show you how to build out of the box today. Not on AWS, and not anywhere. I keep talking about 'figuring it out' in this piece, and that's because this is very (very!) new. I'm excited to get to the "what if" simulation that Animesh was talking about, but we're not there yet. (Or at least I'm not there yet, but I'm doing my best and taking you along with me as I work on this!)

What we can do today is start capturing those decision traces! This is the foundation that everything else seems to layer on to, and AgentCore Memory gives you a neat way to start on this.

AgentCore Memory is hierarchical storage with semantic search, but it isn't a graph database. But! You can (and IMO you should) be smart when you design your namespaces to set yourself up for graph-like query patterns. The SRE agent uses patterns like:

/sre/users/{actorId}/preferences           → user-specific settings
/sre/infrastructure/{actorId}/{sessionId}  → what each agent learned
/sre/investigations/{actorId}/{sessionId}  → decision traces by user

This is smart because it sets you up for success with your query patterns. So if you want to be able to find "every decision that touched Customer X" you could use /decisions/customers/{customerId}. Or, if you want to be able to figure out "all incidents for Service Y"? then you could use /decisions/services/{serviceId}.

Single dimension namespaces are probably enough to get started, and you can add cross referencing down the line as you start to get a clearer pcture of your query patterns. The important bit here is that the decision traces are getting recorded, and then you can build out the graph structure as the tooling (and research) matures.

Once you have figured out how to set up namespaces, it opens the door for you to start searching for precedent before reinventing the wheel and making new decisions:

# Before making a decision, search for similar past episodes
similar_decisions = session.search_long_term_memories(
    namespace_prefix=f"/sre/investigations/{actor_id}",
    query="API latency degradation with memory leak pattern",
    top_k=5
)

# The results include the full episode: situation, intent, assessment, justification
for decision in similar_decisions:
    print(f"Past situation: {decision.situation}")
    print(f"What we decided: {decision.assessment}")
    print(f"Why: {decision.justification}")

This is the feedback loop that unlocks the compounding side of context graphs. You'll know it's all working when previously captured decision traces become searchable precedent for future use, and then every new decision adds another trace.

What's Next

If you want a good place to start digging around, I recommend the Amazon Bedrock AgentCore Samples repo because it has working examples of the patterns I've mentioned in this blog, including the SRE agent that uses memory to personalise investigations based on context from users.

Like I said, we're all still figuring this out! One of the tricky things about writing content about AI is that if I waited until everything was perfectly ready, the blog would feel like it was late. But I really enjoyed reading these pieces, and poking around in the Strands Agents and Bedrock AgentCore docs to piece this together, so hopefully this sparks some ideas for you too. The "what if" simulation capabilities that Animesh describes (where you can ask "what would happen if we changed this policy?") aren't ready yet. But precedent search and pattern extraction are ready for you to try today, and that's enough to start capturing the reasoning that's been missing from so many of our systems for so many years.

The reasoning connecting data to action was never treated as data in the first place, and context graphs are how we might start fixing that. I'm really excited to see this field develop, and if you're working on this please let me know in the comments below!

Once again, thanks to Jaya Gupta, Ashu Garg, and Animesh Koratana for the foundational thinking on context graphs. Their posts are worth reading in full:

AI's trillion-dollar opportunity: Context graphs - Jaya Gupta & Ashu Garg
How to build a context graph — Animesh Koratana

Also, after publishing this post I found the author of the AWS Labs repo with the SRE agent use case that I reference several times in this piece - thank you Amit Arora.

Building a Viral Trend Game in My Terminal with Amazon Q Developer CLI agent

Brooke Jamieson — Thu, 10 Apr 2025 19:24:32 +0000

That viral ice-water-dunking morning routine kept showing up in my feed, and I couldn't help but think it would make a perfect parody in the style of Universal Paperclips - you know, that incremental game where you slowly convert the entire universe into paperclips? So I decided to build a game called "Morning Routine Maximizer" that combines both concepts. And the best part? I built the entire thing without leaving my terminal, thanks to Amazon Q Developer CLI agent.

In this post, I'll walk you through how I created this ridiculous game in one conversation, added some fun hustle culture references, and even got an optimal strategy guide, all from my terminal!

If you're more of a video person not a blog person, see a video version of this post on my LinkedIn!

Starting with Amazon Q Developer CLI agent

If you haven't tried Amazon Q Developer CLI agent yet, it's super handy for developers who feel at home working in the terminal. Unlike previous versions that could just answer questions, this version is agentic, so it can execute commands, write files, and actually get things done for you, all without leaving your terminal workflow.

After installing, I just typed q chat to start.
This opened up a conversation with Amazon Q in my terminal. From there, I asked it to navigate to my project folder:

> go to the morning-routine folder on my desktop please

Amazon Q navigated to the folder using a simple bash command, so I didn’t need to type cd~/Desktop/morning-routine myself!

Describing the Game Concept

Once in the right folder, I described my game concept in detail:

> I want to build a simple Python game called "Morning Routine Maximizer" that parodies both the Universal Paperclips incremental game and viral morning routine videos. 
The game should start with the player as a morning routine enthusiast who clicks a button labeled "Dunk Face in Ice Water" to earn 1 Productivity Point per click. 
The screen should display two main counters: "Face Dunks: X" and "Productivity Points: X". 
Once the player reaches 10 Productivity Points, they unlock the ability to purchase a banana. 
After buying a banana, two new buttons appear: "Eat Banana" and "Rub Face with Peel" (this second button should only become active after eating). 
Eating a banana generates 2 Productivity Points, and rubbing the peel on your face generates 5 Productivity Points. 
As the player accumulates more points, they unlock automation options like "Auto-Dunker" (costs 50 points) which performs one face dunk per second, and "Banana Peeler" (costs 100 points) which processes bananas automatically. 
Later upgrades should include "Multiple Ice Bowls" and "Banana Farm". 
Eventually, players start converting matter around them into ice water and banana peels, tracking this with a "Universe Conversion" percentage. 
The endgame involves converting the entire universe, with victory at 100% conversion. 
Keep the interface minimalist with just text buttons and counters, and include occasional humorous pop-up messages about optimizing productivity. 
This could be implemented using PyGame or even a terminal-based interface, with a simple game loop that tracks resources and updates the display.

My prompt was detailed, covering everything from the initial clicking mechanics to the endgame victory condition. This level of detail is important when you want Amazon Q to generate something specific.

Building the Game in One Conversation

Now - the magic happens! Amazon Q analyzed my requirements and generated a complete PyGame application with all the mechanics I asked for! It:

Created a Python file with over 300 lines of code
Implemented the core mechanics (face dunking for productivity points)
Added resource management (bananas, auto-dunkers, etc.)
Created a proper game loop with increasing difficulty
Added humorous messages about productivity
Generated a README with installation instructions

All of this happened in a single response! The game included everything from buttons for "Dunk Face in Ice Water" to automation systems like "Auto-Dunker" and "Banana Peeler" exactly as I had requested.

Once the files were created, I asked Amazon Q to run the game:

> run the game please

After a quick adjustment (using python3 instead of python- which it picked up on and fixed automatically), the game was up and running on my screen. I had a fully functional incremental game that let me (virtually) dunk my face in ice water to earn those sweet, sweet, Productivity Points!

Adding Some Grindset Humor

The game worked fine, but it didn’t feel fully complete without getting into the grindset with some inside jokes about hustle culture. So I asked:

> Now I want it to be more "in on the joke" with some self aware joke encouragement for "hustle culture" and "grindset" during the game.

Amazon Q didn't disappoint! It updated various parts of the game to incorporate over-the-top hustle culture satire:

Added new messages like "While others sleep, you GRIND! #RiseAndGrind #IceDunkNation"
Changed "Eat Banana" to "Eat Banana Like a CEO"
Replaced "Rub Face with Peel" with "Biohack Face with Peel"
Added stats like "Hustle Level" and "Sigma Grindset: ACTIVATED"
Updated the victory message to include "Your LinkedIn profile now reads: 'Universe Optimization Specialist'"
Renamed the game to "Morning Routine Maximizer: Sigma Edition"

These changes seemed small, but really helped to transform this from a simple game into something that really felt spot on. All these changes were done through natural conversation in my terminal, without me having to edit any code directly, or even open my IDE.

Getting Strategic Advice

Once the game was complete, I was curious about the optimal way to play it. So I asked:

> This is great. Now I want to talk strategy. 
Based on your knowledge of the game mechanics, what are the most optimal first 50 moves in order?

The agent analyzed the game mechanics it had just created and provided a detailed strategy guide, breaking down it’s recommendations into game stage.
Here’s the recommendations:

### Early Game (First 10-15 Moves)
1-10. Manually dunk your face in ice water 10 times to reach 10 Productivity Points (PP)
11. Buy your first banana (costs 10 PP)
12. Eat the banana (+2 PP)
13. Use the banana peel on your face (+5 PP)
14-25. Continue manually dunking your face to accumulate more PP

### Mid-Early Game (Moves 26-35)
26. Buy another banana when you reach 10 PP
27. Eat the banana
28. Use the banana peel
29-34. Continue manually dunking and saving PP

### First Automation (Moves 35-45)
35. Buy an Auto-Dunker once you reach 50 PP
• This is a critical investment as it generates 1 PP per second without manual clicking
36-40. Continue manually dunking while your Auto-Dunker works
41. Buy another banana
42. Eat the banana
43. Use the banana peel
44-45. Continue dunking while saving for your next upgrade

### Scaling Up (Moves 46-50)
46-49. Continue dunking and collecting PP from your Auto-Dunker
50. Buy a Banana Peeler when you reach 100 PP
• This is your next major milestone as it automates banana processing

### Strategy Notes:
• The Auto-Dunker (50 PP) should be your first major purchase as it provides  passive income
• The Banana Peeler (100 PP) is your second priority for automation
• After these first 50 moves, focus on saving for an additional Ice Bowl (200 PP) which multiplies your dunking effectiveness
• Banana Farms (500 PP) come later but are essential for scaling your banana  operations

This strategy maximizes your early PP generation by balancing manual actions with strategic investments in automation, following the classic incremental game pattern of "click to earn, spend to automate, scale to win."

Truly, nothing felt more on-brand for this game than when the strategy recommended prioritizing the Auto-Dunker (50 PP) as the first major purchase since it provides passive income truly in the grindset mindset.

Updating Documentation

To finish the project, I asked Amazon Q to add the strategy guide to the README:

> Can you add this to the README?

It immediately updated the README.md file (did I mention it made a full readme file for me earlier, without me even having to ask?), adding a new "Optimal Strategy Guide" section with all the details from our conversation. The docs now have both basic game information and advanced strategy tips.

The Entire Process in One Terminal Session

What amazes me about this experience is that the entire development process happened in a single conversation within my terminal:

Navigating to the project directory
Describing the game concept
Generating and writing all the code
Running the game
Making substantial enhancements
Getting strategy advice
Updating documentation

I didn't need to switch to my IDE, look up PyGame documentation, or manually edit any files. Everything happened through a conversational interface right in my terminal.

Why This Matters for Developers

This might seem like a silly example, after all, I built a game about dunking your face in ice water and rubbing banana peels on your face, but it demonstrates something really powerful. Amazon Q Developer CLI agent lets you stay in your command line while getting complex tasks done through natural conversation.
For developers who prefer a terminal-centric workflow, this means:

Fewer context switches between terminal, browser, and IDE
Faster implementation of ideas (even weird, trendy ones!)
Complete development cycles without leaving your preferred environment
The ability to make substantial changes through conversation

Whether you're building a meme game or a serious application, being able to go from concept to implementation without leaving your terminal is a huge productivity boost.

Try It Yourself

If you're curious about Amazon Q Developer CLI agent, give it a try for your next project! It's available in both Free and Pro tiers across all AWS regions where Q Developer is supported.

The simple command to get started is just q chat in your terminal after installation. From there, you can navigate directories, generate code, execute commands, and build entire applications—all while staying in your terminal flow.

What 'terminally online' trend should I code next? I'd love to hear your ideas in the comments below!

About the Author: Brooke Jamieson is a Senior Developer Advocate at AWS. They make byte-sized videos with tech tips, speak at conferences all around the world, and write lots of blog posts like this one! You can follow Brooke on LinkedIn, Instagram, TikTok, X, BlueSky and Linktree.

Building a Viral Trend Game in My Terminal with Amazon Q Developer CLI agent

Brooke Jamieson — Thu, 10 Apr 2025 19:24:32 +0000

In this post, I'll walk you through how I created this ridiculous game in one conversation, added some fun hustle culture references, and even got an optimal strategy guide, all from my terminal!

If you're more of a video person not a blog person, see a video version of this post on my LinkedIn!

Starting with Amazon Q Developer CLI agent

After installing, I just typed q chat to start.
This opened up a conversation with Amazon Q in my terminal. From there, I asked it to navigate to my project folder:

> go to the morning-routine folder on my desktop please

Amazon Q navigated to the folder using a simple bash command, so I didn’t need to type cd~/Desktop/morning-routine myself!

Describing the Game Concept

Once in the right folder, I described my game concept in detail:

> I want to build a simple Python game called "Morning Routine Maximizer" that parodies both the Universal Paperclips incremental game and viral morning routine videos. 
The game should start with the player as a morning routine enthusiast who clicks a button labeled "Dunk Face in Ice Water" to earn 1 Productivity Point per click. 
The screen should display two main counters: "Face Dunks: X" and "Productivity Points: X". 
Once the player reaches 10 Productivity Points, they unlock the ability to purchase a banana. 
After buying a banana, two new buttons appear: "Eat Banana" and "Rub Face with Peel" (this second button should only become active after eating). 
Eating a banana generates 2 Productivity Points, and rubbing the peel on your face generates 5 Productivity Points. 
As the player accumulates more points, they unlock automation options like "Auto-Dunker" (costs 50 points) which performs one face dunk per second, and "Banana Peeler" (costs 100 points) which processes bananas automatically. 
Later upgrades should include "Multiple Ice Bowls" and "Banana Farm". 
Eventually, players start converting matter around them into ice water and banana peels, tracking this with a "Universe Conversion" percentage. 
The endgame involves converting the entire universe, with victory at 100% conversion. 
Keep the interface minimalist with just text buttons and counters, and include occasional humorous pop-up messages about optimizing productivity. 
This could be implemented using PyGame or even a terminal-based interface, with a simple game loop that tracks resources and updates the display.

Building the Game in One Conversation

Now - the magic happens! Amazon Q analyzed my requirements and generated a complete PyGame application with all the mechanics I asked for! It:

Created a Python file with over 300 lines of code
Implemented the core mechanics (face dunking for productivity points)
Added resource management (bananas, auto-dunkers, etc.)
Created a proper game loop with increasing difficulty
Added humorous messages about productivity
Generated a README with installation instructions

Once the files were created, I asked Amazon Q to run the game:

> run the game please

Adding Some Grindset Humor

The game worked fine, but it didn’t feel fully complete without getting into the grindset with some inside jokes about hustle culture. So I asked:

> Now I want it to be more "in on the joke" with some self aware joke encouragement for "hustle culture" and "grindset" during the game.

Amazon Q didn't disappoint! It updated various parts of the game to incorporate over-the-top hustle culture satire:

Added new messages like "While others sleep, you GRIND! #RiseAndGrind #IceDunkNation"
Changed "Eat Banana" to "Eat Banana Like a CEO"
Replaced "Rub Face with Peel" with "Biohack Face with Peel"
Added stats like "Hustle Level" and "Sigma Grindset: ACTIVATED"
Updated the victory message to include "Your LinkedIn profile now reads: 'Universe Optimization Specialist'"
Renamed the game to "Morning Routine Maximizer: Sigma Edition"

Getting Strategic Advice

Once the game was complete, I was curious about the optimal way to play it. So I asked:

> This is great. Now I want to talk strategy. 
Based on your knowledge of the game mechanics, what are the most optimal first 50 moves in order?

The agent analyzed the game mechanics it had just created and provided a detailed strategy guide, breaking down it’s recommendations into game stage.
Here’s the recommendations:

### Early Game (First 10-15 Moves)
1-10. Manually dunk your face in ice water 10 times to reach 10 Productivity Points (PP)
11. Buy your first banana (costs 10 PP)
12. Eat the banana (+2 PP)
13. Use the banana peel on your face (+5 PP)
14-25. Continue manually dunking your face to accumulate more PP

### Mid-Early Game (Moves 26-35)
26. Buy another banana when you reach 10 PP
27. Eat the banana
28. Use the banana peel
29-34. Continue manually dunking and saving PP

### First Automation (Moves 35-45)
35. Buy an Auto-Dunker once you reach 50 PP
• This is a critical investment as it generates 1 PP per second without manual clicking
36-40. Continue manually dunking while your Auto-Dunker works
41. Buy another banana
42. Eat the banana
43. Use the banana peel
44-45. Continue dunking while saving for your next upgrade

### Scaling Up (Moves 46-50)
46-49. Continue dunking and collecting PP from your Auto-Dunker
50. Buy a Banana Peeler when you reach 100 PP
• This is your next major milestone as it automates banana processing

### Strategy Notes:
• The Auto-Dunker (50 PP) should be your first major purchase as it provides  passive income
• The Banana Peeler (100 PP) is your second priority for automation
• After these first 50 moves, focus on saving for an additional Ice Bowl (200 PP) which multiplies your dunking effectiveness
• Banana Farms (500 PP) come later but are essential for scaling your banana  operations

This strategy maximizes your early PP generation by balancing manual actions with strategic investments in automation, following the classic incremental game pattern of "click to earn, spend to automate, scale to win."

Updating Documentation

To finish the project, I asked Amazon Q to add the strategy guide to the README:

> Can you add this to the README?

The Entire Process in One Terminal Session

What amazes me about this experience is that the entire development process happened in a single conversation within my terminal:

Navigating to the project directory
Describing the game concept
Generating and writing all the code
Running the game
Making substantial enhancements
Getting strategy advice
Updating documentation

I didn't need to switch to my IDE, look up PyGame documentation, or manually edit any files. Everything happened through a conversational interface right in my terminal.

Why This Matters for Developers

Fewer context switches between terminal, browser, and IDE
Faster implementation of ideas (even weird, trendy ones!)
Complete development cycles without leaving your preferred environment
The ability to make substantial changes through conversation

Whether you're building a meme game or a serious application, being able to go from concept to implementation without leaving your terminal is a huge productivity boost.

Try It Yourself

If you're curious about Amazon Q Developer CLI agent, give it a try for your next project! It's available in both Free and Pro tiers across all AWS regions where Q Developer is supported.

What 'terminally online' trend should I code next? I'd love to hear your ideas in the comments below!

Emoji-Driven Development with Amazon Q Developer CLI agent

Brooke Jamieson — Thu, 10 Apr 2025 19:09:34 +0000

Have you ever tried to explain a complex workflow with just a few pictures? Or found yourself sketching out an application architecture using simple icons? There's something powerful about visual communication - it crosses language barriers and condenses complex ideas into digestible chunks. But would you believe me if I told you that Amazon Q Developer CLI agent is now so intelligent that it can build entire applications from emoji sequences?

This morning, I got to the office and in a pre-coffee daze I started experimenting with what I'm now calling "Emoji-Driven Development" - using of emojis to describe entire application workflows - and I was absolutely blown away by the results. Let me show you what happened!

How Amazon Q Developer CLI agent Takes Understanding to the Next Level

First, a quick refresher: Amazon Q Developer CLI agent launched at the start of March, and is the command line-based version of Amazon Q Developer that goes way beyond just answering questions. Powered by Anthropic's Claude 3.7 Sonnet model via Amazon Bedrock, it can actually execute commands, read and write files, and interact with tools on your system - from package managers to git to the AWS CLI.

What makes it special is that it doesn't just tell you what commands to run - it can run them for you, right there in your terminal. For example - instead of having to say python filename.py to run a file (and then having to go back and change it to python3 filename.py to get it to work) you can just say "run my file" when you're building, and it will decide what command you need, and then run the command for you.

But could it understand something as nonsensical as emojis? That's what I wanted to find out.

The Emoji-Driven Development Experiment

Let's dive in! The concept is simple but powerful: use a sequence of emojis to represent the entire workflow of an application, and see if Amazon Q Developer CLI agent can translate that visual thinking into complete, working code.

I started by launching the agent with a simple q chat command in my terminal, then created five different projects using nothing but emoji sequences to describe what I wanted. Here's what happened:

1. Web Scraper: 🌐 → 📝 → 🔍 → 📊 → 💾

For my first experiment, I prompted Amazon Q Developer CLI agent with:

Create a Python web scraper that does: 🌐 → 📝 → 🔍 → 📊 → 💾

Without hesitation, it generated a requirements.txt with all necessary web scraping libraries, an emoji_web_scraper.py implementing the complete workflow, and a README.md with documentation.
The web scraper followed my emoji pattern:

🌐 Fetch web content from specified URLs
📝 Parse and extract relevant information
🔍 Analyze the extracted data
📊 Visualize the results
💾 Store data in various formats

No additional explanation needed - it just understood what I meant, and got straight to building.

2. ML Image Classifier: 🖼️ → 🔄 → 🧠 → 🏷️ → 📈

Feeling confident, I asked for something more complex:

Write a Python program that implements: 🖼️ → 🔄 → 🧠 → 🏷️ → 📈

Amazon Q Developer CLI agent delivered a complete TensorFlow-based neural network that:

🖼️ Loads and organizes image data
🔄 Preprocesses and augments images
🧠 Trains a neural network model
🏷️ Classifies new images
📈 Evaluates and visualizes results

It even created appropriate documentation and requirements. Almost all of my study notes from when I was first learning AI/ML years ago are full of emojis (can you tell I am a visual learner?) so it’s really nice to see a full machine learning workflow coming together in my command line, just from 5 emojis.

3. Real-Time Data Dashboard: 📡 → ⏱️ → 📊 → 🔔 → 🖥️

Next up:

Create a Python application that: 📡 → ⏱️ → 📊 → 🔔 → 🖥️

This one seems a bit ridiculous now that I’m typing it out again, but it worked really well! This time, Amazon Q Developer CLI agent broke the application into multiple components:

data_fetcher.py (📡) for retrieving data
real_time_processor.py (⏱️) for processing data streams
data_visualizer.py (📊) for creating visualizations
alert_manager.py (🔔) for monitoring and alerts
dashboard_app.py (🖥️) for the web interface

Plus supporting files to tie everything together. It even organized the code into a modular architecture that made perfect sense for a dashboard application!

4. Text Adventure Game: 🗺️ → 👤 → 🎒 → 👾 → 🏆

Time for something fun:

Write a Python text adventure game with: 🗺️ → 👤 → 🎒 → 👾 → 🏆

Amazon Q Developer CLI agent created a complete game architecture:

game_world.py (🗺️) for locations and navigation
character.py (👤) for player attributes and stats
inventory.py (🎒) for items and equipment
encounters.py (👾) for enemies and NPCs
achievements.py (🏆) for quests and rewards

Plus the main game file, docs and requirements. I could immediately run the game and play! The structure followed standard game development patterns - all from just five emojis.

5. Automated File Organizer: 📁 → 🔍 → 📋 → 📂 → ✉️

Finally, I tried:

Develop a Python script for: 📁 → 🔍 → 📋 → 📂 → ✉️

Amazon Q Developer CLI agent created a complete file organization system:

file_scanner.py (📁) for scanning directories
file_analyzer.py (🔍) for analyzing file content
file_categorizer.py (📋) for categorizing files
file_organizer.py (📂) for moving files
notifier.py (✉️) for reports and notifications

Plus a main script, configuration file, and documentation. The system could scan folders, analyze files, categorize them, reorganize them, and then send a report - exactly what the emoji sequence suggested, and exactly what I would normally be too lazy to build myself.

Why Emoji-Driven Development Is Actually Pretty Cool

This isn't just a fun party trick - there are some real advantages to this approach:

Visual clarity: Emojis provide a quick visual representation of complex workflows
Conceptual chunking: Breaking applications into clear, memorable steps
Memorable architecture: Using visual metaphors to understand system components
Consistent structure: Enforcing logical flow from input to output
It's just fun: Adding a bit of joy to the development process never hurts!

What impresses me most is how this showcases the deep understanding that Amazon Q Developer CLI agent has developed. It's not just pattern-matching or following instructions - it's truly comprehending my intent and translating these abstract visual concepts into working code.

The power here isn't just in the novelty - it's about how you can work in your natural workflow and communicate in the way that makes sense to you. Whether that's detailed technical specifications or, apparently, just a string of emojis!

Your Emoji-Driven Development Cheat Sheet

If you want to get started, I've made a reference guide to the most useful emojis for representing common programming concepts:

The key for this is finding emojis that intuitively represent the concepts in your specific domain. Remember, the best emoji sequences tell a clear story about your application's data flow.

Try It Yourself!

Amazon Q Developer CLI agent is available on both Free and Pro tiers in all regions where Amazon Q Developer is available. If you're curious about pushing the boundaries of AI understanding in your dev workflow, I'd love to see what you create!

Let me know in the comments below if you try Emoji-Driven Development - I can't wait to see what emoji sequences you come up with and what applications they turn into!

See you next time,
- Brooke

P.S. For those wondering - yes, I was genuinely surprised this worked so well! I expected to need to provide a lot more context, but it just... got it. The future of development is looking pretty exciting!

The 5 Stages of Using Amazon Q Developer CLI agent

Brooke Jamieson — Wed, 09 Apr 2025 18:21:57 +0000

Have you ever noticed that adopting new tech tools often follows the same emotional evolution as the five stages of grief? Except in reverse - because the end result is pure joy! This pattern became crystal clear to me recently as I watched developers really get stuck in with the new Amazon Q Developer CLI agent.

If you caught my social post about this, you know I've been seeing some fascinating developer psychology happening in real-time. Let's look at each stage of this progression, and figure out what makes the Amazon Q Developer CLI agent such a powerful productivity boost for your everyday terminal workflows!

Stage 1: Denial 😤

"I don't need AI help. My terminal skills are flawless."
Look, I get it. There's something about finally figuring out the command line that gives developers a real sense of pride. Being able to navigate directories, manipulate files, and deploy applications with nothing but keyboard commands feels great, especially when you're working with such a 'minimal' user interface.

When first hearing about Amazon Q Developer CLI agent, heaps of developers react with skepticism: "Another AI assistant? No thanks, I know my Git commands by heart. I can handle my own terminal workflows."

This resistance comes from a good place - years of building muscle memory. But holding onto these habits might be keeping you from an even more efficient workflow.

Stage 2: Anger 😠

"Who needs an AI that runs Git commands? I've been typing 'git status' for 10 years!"

As awareness grows about what the Q Developer CLI agent can actually do, the second stage often emerges: indignation.

"You mean this thing can just take over my terminal? That's MY domain!" There's something almost territorial about it - the command line has been developers' private playground for decades. The idea that an AI can navigate this space feels like an invasion of a hard-earned skill set, and the muscle memory I was talking about in Stage 1.

Pride is a powerful force in tech!

Stage 3: Bargaining 🤔

"Fine, I'll just use it for AWS CLI commands I can't remember..."

This is where the tide starts turning! Developers begin to experiment, usually with very specific constraints:

"I'll only use it for those annoying AWS commands with a million parameters."
"Maybe just for checking my EC2 instances across regions."
"I guess it could help me remember those Kubernetes commands I always have to Google."

This limited testing is crucial - it's the moment developers start to see how the capabilities of Q Developer can slot in with their existing workflow, because it's not something shoved off into a corner anymore.

What makes the CLI agent so helpful is that it's not replacing your skills - it's augmenting them. Behind the scenes, it uses Anthropic's Claude 3.7 Sonnet model for multi-step reasoning, and can use the tools already installed on your system. As Srini Iragavarapu (Director of Generative AI Applications and Developer Experiences at AWS) told me in a recent interview:

"The previous version actually could talk to you. Now the latest version with the CLI is more agentic. What it does mean is not only does it talk with you, it actually talks with the tools that you provide to it through the terminal and get things done for me."

Stage 4: Depression 😢

"I've spent how many hours of my life typing 'aws s3 ls' when I could've been learning guitar?"

Once devs see what the CLI agent can do for them, there's often a moment of existential reflection: "Wait, how much of my life have I wasted on repetitive terminal commands?"

This realization can be a bit sobering. All those hours spent:

Debugging configuration issues
Remembering the exact syntax for complex AWS CLI commands
Running the same sequence of Git operations over and over
Setting up new project environments

I've seen developers become genuinely disappointed when they realize how much time they could have saved. But don't worry - this stage passes quickly!

Stage 5: Acceptance 🎉

"Q, while I grab coffee, could you scaffold a serverless app, configure API Gateway, and write unit tests that validate my Lambda functions?"

This is where the real value becomes apparent! People who get to this stage have fully integrated the Amazon Q Developer CLI agent into their workflow and figured out how to use it consistently, not just once or twice as a little test.

The CLI agent isn't just about saving keystrokes - it's about expanding what's possible in a single terminal session. As Srini shared in our interview:

"A developer finished the task where they thought it would take them three days in less than an hour. Not only did they do that, they wrote tests and were able to get it running as well."

The real power comes from the agent's ability to:

Run commands for you using tools installed on your system
Engage in multi-turn conversations for complex tasks
Access and modify files in your local environment
Query and interact with AWS resources
Use step-by-step reasoning to solve problems

My Personal Adventure

I'll confess - I went through all five stages myself! Even though I create content about AWS tools all the time, I still found myself clinging to my familiar terminal workflow at first.

But once I reached acceptance, I started pushing the boundaries in ways I never expected. Just recently, I conducted what I now call my "Emoji-Driven Development" experiment. I was curious: if the CLI agent is truly understanding intent rather than just following commands, could it build entire applications from nothing but emoji sequences?

So I tried it. I went to my terminal, started a chat with Amazon Q Developer CLI agent, and prompted it with: "Create a Python web scraper that does: 🌐 → 📝 → 🔍 → 📊 → 💾"

I was genuinely stunned when it not only understood what I meant, but immediately generated a complete web scraper that followed the exact workflow I'd intended:

🌐 Fetch web content
📝 Parse and extract information
🔍 Analyze the data
📊 Visualize results
💾 Store everything

This wasn't just a neat trick - I ended up building five different Python applications using nothing but emoji sequences, from ML image classifiers to text adventure games!

I've even built entire games without leaving my terminal. As I wrote in a recent post about this: "This might seem like a silly example—after all, I built a game about dunking your face in ice water and rubbing banana peels on your face—but it demonstrates something really powerful. Amazon Q Developer CLI agent lets you stay in your command line while getting complex tasks done through natural conversation."

That's what makes this CLI agent special - it understands the way we naturally think and communicate, not just execute explicit commands.

Where Are You in all of this?

Which stage are you in right now? Still in denial about needing AI assistance in your terminal? Or have you reached acceptance and started exploring all the creative ways the CLI agent can help in your workflow?

Wherever you are, remember that this emotional path is something I've seen across different experience levels, from beginners to seasoned AWS developers.

The best part about Amazon Q Developer CLI agent is that it meets you where you are. If you just want help remembering AWS CLI parameters, it's there for you. And when you're ready to let it scaffold entire applications while you focus on higher-level architecture decisions, it can do that too.

Want to get started with the free tier of Amazon Q Developer and begin your own exploration? Check it out here.

Let me know in the comments which stage you're in, and what your experience has been with the Amazon Q Developer CLI agent so far!

How I became an AWS Machine Learning Hero

Brooke Jamieson — Tue, 14 Jun 2022 05:04:16 +0000

A little over 2 years ago, I set myself the goal of becoming an AWS Machine Learning Hero, and it has finally happened! This was a really important step in my career journey, and in this post I'll share a little bit of my pathway including the public-facing parts, and the behind-the-scenes elements.

According to the AWS Website, "the purpose of the AWS Heroes program is to recognize and honor the most engaged and influential developers who have a significant impact within the community". This is a broad description, but if you've ever seen AWS Heroes presenting at events or in online content, it should make a lot more sense! There's a really broad range of builders from around the world with different skills and experience, but overall it's a really impressive bunch and I can't believe I'm included in this!

There's a range of categories including Community Heroes, Container Heroes, Data Heroes, DevTools Heroes, IoT Heroes, Machine Learning Heroes and Serverless Heroes and these categories contain conference speakers, open source contributors, meetup organisers, authors, technical trainers and content creators.

At the time of writing this article, I'm one of just 34 AWS Machine Learning Heroes in the world, and by my count there are just 3 Australians including me in this list (😱). It's a huge honour to be included in this list, and hopefully this article will help some of you to make it on the list as well!

How did I get here?

Short answer: Working really, really hard!
Longer answer: The three main parts are my time as an AWS Community Builder, my content creation, and my global conference speaking experience.

Important note - This is just a guess based on my personal experience and I'm not in the room where it happens so I can't be 100% sure. The official website says some key characteristics of an AWS Hero include:

Enthusiasm – An individual who regularly participates and engages the AWS community.
Expertise – Being an expert in AWS and staying on top of trends.
Leadership – Building relationships and strengthening ties within the community.

What worked for me might not work for you, but hopefully it's a start!

AWS Community Builder

The AWS Community Builders program is fantastic, and was a really key factor in my personal and professional development over the last few years. The program is application only, and if applications are closed when you read this I really recommend joining the wait list. If you're interested in joining, I also recommend reading Stephen Sennett's article 'How to Become an AWS Community Builder' for a thorough overview!

I was an AWS CB for around a year, and the key piece of advice I'd add to this article is that the program is called "AWS Community Builder" and not "Best AWS Builder" or "AWS Secret Builder" - it's not just about being good at cloud computing, it's about sharing this information and getting involved with your community in person and/or online. Don't let a lack of perceived technical ability or confidence stop you from applying, but also don't forget that if you are really technically proficient, this might not be enough for your application.

TL;DR there's no point getting the right answer if you can't tell anyone about it, and so being an AWS Community Builder does involve the community!

Content Creation

Being a developer is all about being able to learn new things on the fly, but most of us put off learning how to write Minimum Viable Blog Posts for far too long. I had been creating lots of content, both for myself and ghost writing for others, before I was in the AWS Community Builders program but it was a nice encouragement to help me accelerate output.

I publish content on my dev.to, on my LinkedIn, my Twitter and most recently on TikTok. This is probably a good time to remind you that this isn't what is stipulated as a requirement to become an AWS Hero, and it's just what I happened to do!

Also, it's important to note that I wasn't making this content because I had to, or because it was part of my day job - it's just what I liked to do! If you've watched my TikToks you'll know that I just genuinely enjoy this, and it's a really good way for me to connect with my community. I have even made two AWS themed aumented reality TikTok filters - including 'Which AWS Service Are you?' and this AWS Summit Crown.
Did anyone (at all!) ask me to make these? No
Did they have anything to do with me becoming an AWS Hero? Probably not
Did I have a great time making them? Absolutely!

Conference Speaking

Louder for people in the back: Please don't count yourself out of conference speaking if you are neurodiverse or if you have anxiety!
I can't stress this enough, but I'm not an outgoing or confident person by nature. And yet, I've done over 40 conference talks and speaking engagements over the last few years. Candidly, it was a multi year journey for me to just attend an event without having a panic attack, before I got the courage to ask questions as an audience member and then start to deliver my own talks. It wasn't easy, but it was so worthwhile!
This is very meta, but if you are interested in getting started as a conference speaker, I have an hour long conference talk freely available on Youtube from NDC Sydney called 'How to Become a Tech Conference Speaker' where I detail the entire process:

Speaking at conferences is a great way to improve your communication skills, but it's also a really good way to get more involved in your community! Conferences normally provide all-access passes to speakers, and so I've been able to attend so many fantastic events without needing to pay for tickets. I've learnt so much from other sessions I've attended, but I've also made so many friends in the community who I otherwise might not have met. I've also gotten to travel internationally to speak at events, and this isn't something I ever thought I'd be capable of.
It's difficult to start, but I promise it's worth it!

Bringing it all Together

The key theme is that it seems to be about being a leader in the AWS Community, not just an elite dev in private. It's about creating high quality content and making a positive impact on the global AWS Community. Get involved with AWS Community Builders, and with the Community & Developer Advocate teams in your region or area of expertise. Start attending and presenting at your local AWS User Group and overall, just enjoy yourself.

There's no one-size-fits-all solution, but this is what worked for me! If you have any questions, pop them in a comment below!

About the Author: Brooke Jamieson is an AWS Machine Learning Hero from Brisbane, Australia. Learn more about their journey as an AWS Developer, watch this video, or view their profiles on LinkedIn, Twitter, Dev.to and TikTok.

Everything I've learned in 3 months on TikTok

Brooke Jamieson — Mon, 24 Jan 2022 23:14:58 +0000

So I'm not new to social media, but I put off starting a TikTok for a long time because I just couldn't handle the idea of another platform to maintain. I already have a LinkedIn (with articles like this one!), a fashion Instagram (a souvenir from my past life as an international model), a Twitter and a Dev.to for my tech writing. However, I finally bit the bullet at the end of October and started a TikTok, and I have learnt so much along the way! In this article I'll outline everything I learnt the hard way about the platform, and all of the benefits I hadn't considered before making my account.

I can already hear you groaning about TikTok - I used to think it was just a dancing app for kids too!

I challenge you to put your existing reservations aside just for this article, as I really think there are huge gains to be made via this platform for the overarching tech landscape and community.

1. Growth is hard, but just push through

There's no shortage of experts making videos about how to grow your account to XYZ followers in XYZ days and it's important to remind yourself that if all of their tips worked, none of these accounts would exist. Everyone is trying to figure things out at the same time, and some advice is great but other advice is complete garbage - instead of trying to find magic beans just find a consistent cadence to upload that works for you, and figure the rest out as you go along.

Here's my last 60 days worth of analytics - note that things really fluctuate depending on if I uploaded anything, and if it caught on:

On the other hand, this is my cumulative follower count:

Things are pretty stagnant at the start because I ended up taking 10 days break, but then things really picked up in the holidays when I had more time (and mental energy!) to post.

After consistent posting, TikTok isn't the largest platform I have (yet!) but it is my fastest growing.

2. So many users want to transition to tech careers

This is by far the biggest realisation I had - the user base is fundamentally different to all of my other networks, and there are thousands of young people wanting to explore tech careers, as well as established professionals who want to transition to a tech career.

I think this builds on the narratives around 'The Great Resignation' or as Kara Swisher puts it - 'The Great Reassessment'. So many people have seen glimpses of tech careers, either through 'Women in STEM' movements or just more transparency around the people behind the tech platforms they use on a daily basis, but they still have absolutely no idea what sorts of jobs are available in tech. There's a fantastic community of creators who are pulling back the curtain on what it's like to work and flourish in tech careers, and there's so many interested viewers!

I'm yet to see organisations really make the most of this at scale - there's lots of brands using TikTok for B2C marketing (Duolingo is my absolute favourite example of this - it's absolutely unhinged and chaotic which is perfect) but there's not a united push by large organisations to use the platform for employer branding/recruitment/upskilling of the incoming talent pipeline. This is a huge opportunity, which would benefit the industry at large.

3. Don't put effort into content

The message is so much more important than the method! There's always a lot of talk about what "the algorithm" prefers but at the end of the day, humans are human! Relatable stories, or easy-to-digest messages will always do better than well-produced nonsense. Just as there are so many business books that should have just been a blog post, there are many blog posts/ebooks that would have much higher engagement and recall if they were distilled down to something short and memorable.

You don't need Final Cut Pro/Premiere Pro, you can edit as you go in the app and it's faster than you think.

4. The videos you don't like will end up doing the best

Similar to point 3 - there are so many videos I put effort into that I was proud of and just didn't end up doing well at all. Big examples of this are my first two videos - this one about AWS Snowball and this one about Amazon SageMaker. They're absolute nonsense (no one will dress up as SageMaker for halloween) but I made these before I worked out how to use the TikTok native editor so they took so long to make and they didn't get any traction! Another example is this one-take-wonder of me reciting Julia Gillard's misogyny speech. I love this video but it didn't really catch on - perhaps my viewership is too young to remember this!

On the other hand- the videos I absolutely hate are my best performing videos. Eg this one which has 44100+ views and this one which has 27500+ views. These were my laziest ideas and I put zero thought or energy into them, but they did really well.

With that being said - videos "doing well" is not the end game! Getting to the fyp (the 'home' screen with lots of videos) for other users is great to reach new audiences, but there's no point doing this if there's nothing of substance to keep the new audience engaged with your profile. Like most things in life, it's all about balance.

6. The comments are ruthless

As someone who has received more than their fare share of revolting instagram DMs (and occasionally revolting LinkedIn Messages) I thought I'd be fine with TikTok comments - I was wrong! The comments are as brutal as any other platform I've been on which I wasn't expecting at all. The 'rude dudes' are a vocal minority and they are wildly outnumbered by wholesome and supportive users, but they exist nonetheless.

If your kids are posting on TikTok definitely check in with them and make sure they're ok!

7. The key to "The Algorithm" is UX

There's always so much talk about what "The Algorithm" prefers or pushes out to users, but the backbone of "The Algorithm" is a really interesting UX strategy.
On the 'for you page' where you can swipe through a stream of videos, there's only ever one video shown on screen at a given time. As a result, any user signals (eg how long someone watches a video, if they replay it, if they like/comment/share the video or if they click through to view more videos with the same audio or more videos from the same creator) can be directly tied back to that content.

This strategy is so simple but so effective!

It's very different to the 'explore' page on Instagram which can show ~17 pieces of content at the same time. If a user dwells on this page, there's no quick way to understand what they're actually pausing to look at, so it's impossible to tie this back to a reinforcement learning model to improve "The Algorithm".

This is something I talk with Blackbook.ai clients quite often. It's not about having the "fanciest" algorithm - it's about having good data to work with. For some clients this means modernising and de-risking their Data Warehouse/Data Lake and for other clients, this means optimising internal processes and UX of internal products to make it easy and intuitive to give feedback signals and improve the effectiveness of an AI solution.

TikTok is the best example of this I've seen 'in the wild' and it's a valuable take-home message even if you never intend on actually using the platform.

About the Author: Brooke Jamieson is the Head of Enablement - AI/ML and Data at Blackbook.ai, an Australian consulting firm specialising in AI, Automation, DataOps and Digital.
Learn more about Blackbook.ai here and learn more about Brooke here.

2021 AWS re:Invent Recap Video with Mike Chambers and Brooke Jamieson

Brooke Jamieson — Sun, 05 Dec 2021 22:47:27 +0000

Thanks so much to Mike Chambers (AWS ML Hero)for inviting me to chat about key takeaways from AWS re:Invent 2021!
We both live in Brisbane, Australia so we were both unable to attend re:Invent in Las Vegas in person, but we still found a way to experience the community aspect of the event!

Watch the video here:

You can follow Mike on Twitter here: https://twitter.com/mikegchambers

And if you're interested in learning more about Machine Learning on AWS, I really recommend Mike's course for the AWS Machine Learning Specialty Certification aws-mls-c01 Exam!

Transcript:
Mike: Hello everybody, well Reinvent 2021 is still going on in Vegas. I think around about now it's party time in Vegas and they're all partying it up at replay. But for some of us who aren't able to be there, well, we can't be there. But reinvent for me is all about meeting people and connecting with the community. That's why I go to reinvent. And so I'm doing that in my town and I was really grateful to find Brooke who is community builder for AWS in Brisbane, my hometown. Hey Brooke!
Brooke: Yeah hi, thanks for you having me! I would add that it's not just that people are partying after reinvent, now they're probably all starting to get sick, not with the panini, but with probably any other thing that they've come down with from partying too hard. So thoughts for everyone currently struggling if you’re watching this
Mike: Fatigue? Yes, yeah. So this is the first time we've met, like we've had like literally two minutes before I turn the camera on. But yeah, it's a pleasure to meet you. I like walking down the corridors at reinvent, bumping into people, and so this is what we've done virtually. Real really kind of, virtually speaking, so how's reinvent? Been for you on this side of the world?
Brooke: Yeah, very chaotic. I'm no stranger to bad time zone differences. I was actually supposed to move to America in April 2020. So I worked for most of 2020 in both Brisbane and New York Time zones, which were about 14 hours difference, so I knew ahead of time that I wouldn't be able to do that this week 'cause I had so much of my actual normal day job to do, so I'm Head of Enablement for AI, Machine Learning and Data at Blackbook.ai , which is an AWS Partner as well. So doing that all around everything. Constantly jealous not only of everyone having a wonderful time at reinvent, but all of the time that they had to actually dedicate to watching things. I think I'll be catching up for about a month after this at this rate.
Mike: Yes, yeah. Well so for me, going to reinvent. I've said this before. I don't think this is a secret I've never been to a reinvent session. Any of the reinvents I've been to. Have you been to reinvent?
Brooke: I have not been to any reinvent sessions or any reinvents.
Mike: So well, having been there, I've not been to any other sessions either. I would just catch up with them afterwards. Online on demand, even pre covid times, but it is about the people and yeah no so obviously. Hopefully next year we can both be there. Yeah, and connecting with those people. So machine learning as you said, so we both sort of share that common interest in machine learning. Anything in particular that sort of caught your eye in relation to a machine learning at reinvent this year?
Brooke: Lots of different applications of things. I work in very Applied AI stances, so although I studied pure maths at uni but now it's very very applied. So how does this connect to industry? And I think as part of the maturity of lots of the services, the way they're tying in with different things is really good. So for example, the chat bot builder for Connect. Connect which is my third favourite AWS product. So now being able to use those call transcripts to make a chat bot because you could always make transcripts and you could always make a chat bot, but for many organisations that leap was just. Too hard, so that's really cool, and even things like amplify the front-end tools, not a front end developer, but I know a lot about Figma so I can see that getting used at scale and really large organisations as well.
Mike: That seems to me almost like a bit of a theme reinvent was wrapping easier tools on top of existing services, so making, lowering the barrier to entry for a whole bunch of things in tech.
Brooke: Yeah, it's a different view of accessibility, so obviously accessibility for actual accessibility is really important, but I think. As the underlying tools are getting more mature, people are now able to ask those applied questions about how does this actually fit in with what we're doing? Or how can I actually use this? Because especially for pretty much all of the AI and machine learning services, they're very good back end tools, but it's not as tangible as something you can just work out yourself. So I think having friendlier front ends for things is a really good trend to see happening, and that's something that I'm hoping will continue as everything continues to just mature, and I imagine lots of this serverless aspects of things will help, 'cause that's a really big maturity step as well. So I think yeah, especially next year after that it'll be even.
Mike: Sure a couple of things from what you just said there. That sort of tie directly into announcements, so different frontends, uncomplicated tools. So canvas. Canvas is, I think, as we've discovered on live streams on this channel before as well, live Canvas is a different front end onto SageMaker studio, like it is the same tool. That's great 'cause it really does sort of drop that barrier to entry. Makes it possible for someone who's not a machine learning centric person, but is a domain expert in data. I would imagine you'd need to know what the data is you're looking at, but can make it easier for you to train models.
Brooke: Yeah, and I always talk about context and I think that's the most important part of applying machine learning to business things. Is understanding all those contextual elements and how it all fits together. Even I think you can add automatically. I saw in Allie’s [Allie Miller] part of the keynote about adding the holidays in. That's really good context that before people just wouldn't have thought of because it would have been too hard. Yes, I think that part of Canvas is really good. Also, I was joking on Twitter, but um, the illustrations they have of the girl and the things I like her outfits and I aspire to make I don't know. Hopefully that will be a side merch line of just I can go in Halloween costume is the canvas girl next year.
Mike: This is the Etsy store that you're going to set up, right?
Brooke: Yes, OK, manifesting itself in jumpsuits so I can feel like the canvas girl!
Mike: So something else you mentioned just in that bit before as well, was just a serverless bit, and they announced a serverless inference end points for its age maker, but they almost set it as though it didn't really matter, and they kind of just moved on. Yeah, that to me was really exciting. What is that? Something that you think you might using? You do a lot of the practical business side and actually seeing the rubber hit the road in terms of machine learning as well. So yeah, do you see a big use case for that?
Brooke: Yeah, definitely, especially when that ties into all of the sustainability objectives. I think as part of I don't know, the serverless part is good for people to actually using it, but when you look at how that underpins all of the other sustainability pieces of the reinvent puzzle, I think that's going to be really helpful as well. Just now, people are really starting to strive for that more, and it's getting reported on more Boards and Company Director levels, so anything that enables that process alongside other technical maturity in the organisation will really help them.
Mike: Not machine learning, but well, not directly machine learning, but another thing that was announced, just very recently was the. At sixth pillar of the well architected framework, which is sustainability which I just have to mention. Obviously because you just talked about serverless in the sustainability effort there. I’m super excited about this. 'cause the idea that there is now this wealth of information that will support businesses to actually have real tangible conversations about sustainability rather than just saying yes, we think we're gonna do the right thing in 20 years time. There's something that people can actually do now.
Brooke: Yeah, and I started. I've been researching this heavily for the last six months or so and I started working out how I could manually. Calculate a lot of these things and then I saw that reinvent was coming up and I was just like crossing my fingers. That 'cause I knew I was like oh so I can get this from this and this from this and plug it together. And then I was like hang on, there's going to be some sort of announcement I saw there was a state sustainability guide for reinvent. And then I was just hoping fingers crossed that they would release the more mature features. And I got so much more than that with their well architected pillar. So I think that's good because it will force people to think about it and take it into account. Even so, because even if they're not trying to improve things, you can't improve on something you don't understand the baseline of. So I think a huge piece of this will just be people understanding where they actually are sitting as a baseline, because there's no coverage of that at so many different organisations.
Mike: Yeah, we're really supporting that, and the other part to that. Just before was it the day before? I think it was in Peter’s Keynote was the Dashboarding and the metrics around carbon emissions. All that kind of stuff. And that goes to your point as well. That just supports you. And I was just having a conversation just before I came here and. This will be up on the channel at some point as well with Luke Hargreaves, who was one of the principal architects behind actually putting together a lot of this stuff and so that was a really interesting conversation. Part of that was I think it was my suggestion and if it wasn't I'm going to take it as my suggestion that we actually develop like a little open source widget that can go on companies websites that can hook into those metrics so that they can just be here we go. This is this is how good or bad we are. That would be cool.
Brooke: Yeah, if you've ever used like in your going Chrome developer tools and you can look at the lighthouse audit for websites, it gives you four different scores that are traffic lights coloured as well as one to 100 for performance and accessibility and SEO things like that. But it's really good. Very assigned to this, but it's really good if you are running for example in E Commerce store and you want to see what your competitors stores are like relative to yours in terms of performance, but it's just in Chrome developer tools and you can look at how everything is working and the idea that it's sort of scores you on these. Different levels, something like that, I think would be really cool. Just because it not only gives a baseline for you and then you can map any improvements you have, but you can also then see how you're comparing to others, and I think. People are really fuelled by competitiveness, even if they're not fuelled by just being better for the planet. So I'll take whatever I can get for this point. Whatever pushes it forward. Yeah, absolutely.
Mike: Alright, so so. Just before we started rolling here, we were just talking about reinvent what we took away from it. You have an additional thing that you think that people can sort of take away from reinvent.
Brooke: Yeah, yeah, so I do lots of speaking about speaking, which is incredibly funny when I say it like that, but lots of things about how to present your data so it's persuasive. How to make your voice heard just because as a young woman in tech, that's not always easy out of the gate. So I try to make it easier for other people. And I mentioned this in my reinvent session guide, but I think it would be really helpful for anyone in tech. Once you've had your watch through of all the keynotes, the exciting time when you're excited about the new features, watch it again through the lens of a business person and see how they're making the points, see how they're communicating things, because it is an absolute masterclass in tech communication. I don't know how big the teams are that are writing these talks, but it's not Swami writing his own machine learning keynote.
Mike: I’m sure he’s involved, but probably there's quite a lot of other people.
Brooke: There would be a lot of people helping him with that, and obviously it would be fantastic if I had that every time I do a talk, but that's not the case, so make note of how they're making points, even how they transition between some of the business, technical case studies and then the technical underpinnings of that. You won't get a better overview of that in action, so I think it's really valuable to you if you can really work out how that's actually happening, and look at how they're communicating and even looking at what's on their slides and how they communicate those slides. It's one thing to take that away from more business presentations, but they're presenting about AWS services. They're presenting about data, they're showing data visualisation. Look at that and see what they've actually done and what stands out to you so that you can use that. Or take what you want from it next time you're doing a talk or next time you're presenting to the board, and I think it would just help so much to be able to communicate lots of this to business people, whether that's for if you're trying to sell them something, or if you're wanting to be able to transition to one of these new services internally and you need buy-in, fantastic there is someone that's showing you how to communicate that to a business audience using benefits that will make sense to them. So think about how you can use that. Communication piece,
Mike: I like that a lot, I actually watched on an inconvenient truth, which was many many, many years ago now and strangely on topic still. But that was one of the things that I sort of wanted to take away from that I actually demonstrably changed my PowerPoint presentations because of watching that, and they did also have a massive team of people putting that presentation together. I think that worked from Apple. They had a big keynote steep team working on it, but just don't put. Loads and loads of bullet points in your presentations, but also to your point. Listen to the way they were talking about. Actually, the way that they sliced up the message is that?
Brooke: yeah, you can transcribe them. Obviously, in AWS Transcribe or if you are lazy or efficient, you can just in YouTube. You can open up the transcript to the side and just copy paste it and even looking at different you can do text analytics on it. I went through as a. Like a starting point, I just count how many times they talk about various things. So how they talk about how many times they bring up data? Or you can get a lot of sense of what's most popular overall in a talk. If you're looking at different service names or when they come up, but then even if there is something in the transcript that you think is a really powerful way of communicating, you have it in the transcript so you can clip it out and put it in notes or something and it will. Just it's a weird way to interact with content if you're not used to it, but it's really helpful in working out how you can level up from it. Just because if you tried to watch a course about how to present technical content. There's no way they would be able to put everything in that was specific to AWS. Sure, it would make sense, but just approach it like that and then get more value out of it. That way.
Mike: I like that. So go watch all the presentations that you want to watch and learn all about the services you want to watch. But then consider going back and looking about it again and sort of getting that extra layer detail from it. I really like that.
Brooke: Especially as well. Just as another side note, if there's ever any of the workshops, or if they put any of the other ones on that level 100, so it's normally the really basic level. Go and watch those as well. As if it's through the lens of someone you manage or someone that's new to tech and then look at how they present those topics as well. Because then next time someone comes to you and says where do I get started with machine learning or something like that? You'll understand the direction that they step through those points and how they sort of package things together and also where they stop. Think when you get really in deep with SageMaker and things like that. It can be hard to look at what's actually included in a beginners level. So yeah, I would say put your business hat on and then put your either entry level developer or someone that you manage. Put their hat on as well and then re watch some of the basic content.
Mike: Try and remember what it was like for you when you first started. Really try 'cause like there was that day when someone said I haven't. I haven't notebook and you go. What do you talking about and then then everything from there started to unfold but just try and put yourself back in the shoes if you before you knew any of that stuff it's hard to do. It's hard to do.
Brooke: Or the first time you heard the word container. Yeah, yeah.
Mike: Or Serverless, or any of these things?
Brooke: Absolutely yeah. So look at it through those lenses because they honestly the teams that it'll be us have been working on this for so long. And then the really big Sprint to the end. So I think you'll get lots more out of it. More broadly for your career than just the technical underpinnings.
Mike: Excellent thank you for that. So Brooke if people want to find you and connect with you and do all the socially kinds of things, where should they go?
Brooke: It depends what you want! So I have a Twitter that's just my name Brooke Jameson that's mostly AWS tech content. I was supposed not on LinkedIn about anytime I'm going to be speaking either in person or virtual events. So that's my speaking stuff over my LinkedIn. I also started at TikTok and if I talk about it here that means I have to keep posting on it, but my TikTok is all about how to transition into tech and getting a job with AWS or getting a job using AWS or just getting more into cloud for career purposes. That's sort of yeah, how they split up.
Mike: Well, thank you very much Brooke. I'm glad I bumped into you reinvent, even though we're not actually there
Brooke: Our fake reinvent!
Mike: In a very, very quiet room compared to anything that you might find in Vegas. But thank you so much!
Brooke: Thanks. Maybe you'll find a 10-year jacket under the table or something.

About the Author: Brooke Jamieson is the Head of Enablement - AI/ML and Data at Blackbook.ai, an Australian consulting firm specialising in AI, Automation, DataOps and Digital.
Learn more about Blackbook.ai here and learn more about Brooke here.

AWS IoT TwinMaker: The Hidden Gem of re:Invent

Brooke Jamieson — Wed, 01 Dec 2021 06:31:35 +0000

Overnight, AWS made lots of product announcements at re:Invent 2021, and while lots of these are getting good coverage, I wanted to quickly bring everyone up to speed on what I think is the hidden gem of re:Invent - AWS IoT TwinMaker.

Before I started as Head of Enablement for AI Machine Learning and Data at Blackbook.ai and before I became an AWS Community Builder for Machine Learning, I was actually User Experience Lead and then Analytics Lead at a global IoT Proptech platform, so I've actually got heaps of experience with digital twins for facilities optimisation, and I've seen a lot when it comes to integrating disparate data sources in the built environment.

This is why I'm so excited about AWS IoT TwinMaker which is a new service which makes it faster and easier for developers to create and use digital twins of real-world systems to monitor and optimise operations.

Diagram from the AWS IoT TwinMaker Product Page:

Data from sources like equipment sensors, video cameras, and business applications is being collected and processed by all sorts of organisations about their equipment and facilities. In order to optimise industrial operations, improve production output, and improve equipment performance, most companies want to combine these disparate data sets to create a digital replica of their physical systems. This is called a Digital Twin. Although creating and maintaining Digital Twins may seem difficult and time-consuming, it doesn't have to be anymore.
With AWS IoT TwinMaker, you can quickly get started with creating digital twins of equipment, processes, and facilities. This is done by connecting data from different data sources like time series sensor data from equipment, video feeds from cameras, and maintenance histories from business applications. Importantly, You won't have to move the data into a single repository for this to work.

Building a Digital Twin Graph allows your operators to make better data-driven decisions by understanding and connecting all the sources of your data into a 3D application of the physical environment that displays data and insights in a spatial context.

There's built-in data connectors for services like AWS IoT SiteWise for equipment and time-series sensor data, Amazon Kinesis Video Streams for video data, and Amazon S3 for storage of visual resources like CAD files and data from business applications.

AWS IoT TwinMaker forms a digital twin graph that combines and understands the relationships between virtual representations of your physical systems and connected data sources, so you can intuitively and accurately model what's happening in your real-world environment.

Once you've built the digital twin graph, you can visualise the data using the context of the physical data using your existing CAD files or point cloud scans to make 3D models of your space.

Example from Github repo of AWS IoT TwinMaker Samples:

From here, you can then overlay video and sensor data overlays from your connected data sources as well as insights from connected machine learning or simulation services. But it's not just forecasts and projections, you can also add practical elements like equipment maintenance records and manuals to the digital twin which becomes your single source of truth.

AWS IoT TwinMaker is available today in preview in US East, US West, Europe and Asia Pacific, and this is again just a preview service so this list will grow. The AWS Free tier will include 50 million API calls per month for 12 months, so I'm excited to get building with this!

About the Author: Brooke Jamieson is the Head of Enablement - AI/ML and Data at Blackbook.ai, an Australian consulting firm specialising in AI, Automation, DataOps and Digital. Learn more about Blackbook.ai here and learn more about Brooke here.