Last December, I delivered a talk on AWS Bedrock at AWS Community Day. The main purpose of this talk is to differentiate between traditional AI and Generative AI, and how it makes a difference to the world, and AWS Bedrock enables us to use different models with different capabilities.
First, I explain generative AI and now how almost in every application Generative AI is used and the common challenges a developer face while building an application where they need more then one model for different capabilities as they jump from one plateform to other for using other models as different models have diffenent capabilities For example, for coding developer preffer Anthropic modes, for explaination open AI models work amazing and many other models for different capabilites.
AWS Bedrock solves this problem by offering multiple models from different companies.
I attach some of the companies' models that AWS Bedrock offers :
There are a couple of reasons developers should use AWS Bedrock :
● The developer doesn’t have to jump from one platform to another for different models
● They don’t have to worry about the infrastructure; AWS manages all for you.
● The best part, which I like, is that your data is not being used for model training, so you can use different models with ease of mind.
Now the question is how developers can use these models in their application or test the capabilities, Soo mainly there are three ways to use or test the model:
● AWS CLI
● Bedrock API
● Playground
How AWS Bedrock works. Here is the simplest workflow:
User Prompt
Amazon Bedrock
AI Model
Generated Response
Most developers use the API Call method, where they directly use the model by API Call:
Now the next Question is how to choose the right model, it all depends on the use case and requirements:
● Model Type & Multimodal Capabilities
● Performance & Latency Requirements
● Model Size & Context Window
● Customization Needs
● Licensing & Commercial Use
If you have any queries about AWS Bedrock, let me know in the comments.
Top comments (0)