All Data and AI Weekly #245-08 June 2026

All Data and AI Weekly #245-08 June 2026

( AI, Data, Agentic AI, Cortex Code, NiFi, Iceberg, Polaris, Streamlit, Python, Java, SQL, MCP, LLM, RAG, Cortex AI, AIFN, Search, Unstructured Data )

A Couple Quick New Links

• https://github.com/tspannhw/nifihub

• https://docs.snowflake.com/en/LIMITEDACCESS/remote-development-vscode-extension

• https://github.com/typpo/textbelt

• https://www.snowflake.com/en/dev-day/americas-virtual/

• https://github.com/zdk/lowfat

• https://github.com/microsoft/pg_durable

• https://medium.com/@matiasmaquieira96/snowflake-dcm-projects-a-quiet-revolution-in-dataops-7ae39019d68f

• https://docs.snowflake.com/en/user-guide/snowflake-cortex/pricing

• https://docs.snowflake.com/en/developer-guide/snowflake-app-runtime/about-snowflake-app-runtime

• https://docs.snowflake.com/en/release-notes/2026/10_20

• https://github.com/Agent-Field/sec-af

• https://github.com/anthropics/defending-code-reference-harness

• https://snowflakechronicles.medium.com/your-schema-has-too-many-views-virtual-columns-can-fix-that-ca81cb2be855

• https://medium.com/@obelix74/a-30-million-request-load-test-for-apache-polaris-fb5690040154

• https://medium.com/@chandra.nayak_2494/connect-salesforce-agentforce-to-snowflake-data-in-minutes-with-mcp-61edf3d2b9b8

My Talk at Summmit

https://github.com/tspannhw/SNACKAI-CoCo-TrafficVideo

---

Early Summary Blog

https://medium.com/@luke.turanski/snowflake-summit-2026-early-recap-and-first-impressions-19178eae675d

Featured: Snowflake to Acquire Natoma - Governed Agentic Access

Snowflake announces intent to acquire Natoma, bringing secure connectivity and governed agentic access to the enterprise. Natoma provides the secure connectivity layer that AI agents need to safely interact with enterprise systems.

Blog: Snowflake to Acquire Natoma for Governed Agentic Access

Press Release: Snowflake Announces Intent to Acquire Natoma

Cortex AI & Agents

Gemini 3.5 Flash for Multimodal Video and Audio (May 28)

Gemini 3.5 Flash is now available in Cortex AI Functions for multimodal video and audio analysis. Frontier multimodal intelligence with Flash-level speed for marketing, content intelligence, compliance, and media analytics.

Release Notes: https://docs.snowflake.com/en/release-notes/2026/other/2026-05-28-gemini-3-5-flash-multimodal-video-audio

Documentation: https://docs.snowflake.com/en/user-guide/snowflake-cortex/ai-multimodal

Claude Opus 4.8

Anthropic releases Claude Opus 4.8 - the latest flagship model with enhanced reasoning and coding capabilities.

Read more: https://www.anthropic.com/news/claude-opus-4-8

AI_EXTRACT Extraction Scores (GA - May 22)

Extraction scores for AI_EXTRACT are now GA. Get confidence scores (0-1) for each extracted value, with support for entities, lists, and tables.

Release Notes: https://docs.snowflake.com/en/release-notes/2026/other/2026-05-22-ai-extract-scores-ga

Documentation: https://docs.snowflake.com/en/sql-reference/functions/ai_extract#extraction-scores

AI_COMPLETE with Fine-Tuned LLMs

Example repository for using AI_COMPLETE with fine-tuned LLMs on Snowflake.

GitHub: https://github.com/sfc-gh-csharkey/AI_COMPLETE_w_FineTunned_LLM

CoCo Control Hub: Self-Service Cortex Code Spend Governance

CoCo Control Hub - a self-service Cortex Code spend governance app built on Streamlit in Snowflake (SiS).

Read more: https://medium.com/@ramkumar.c233/coco-control-hub-a-self-service-cortex-code-spend-governance-app-built-on-sis-f411dfe59ee9

Snowflake Summit 2026: Complete Announcements Report

The Agentic Enterprise — San Francisco, June 1–4, 2026

Report Date: June 8, 2026
Event: Snowflake Summit 26 — Moscone Center, San Francisco

Attendees: 20,000+ in-person, thousands more virtual

Total New Capabilities: 26+

Executive Summary

Snowflake Summit 26 (June 1–4, 2026) was the largest user conference in Snowflake's history, bringing more than 20,000 in-person attendees to the Moscone Center in San Francisco. Under the banner of the "Agentic Enterprise," Snowflake unveiled 26+ new capabilities across six domains designed to transform the AI Data Cloud into a full control plane for enterprise AI.

The summit's headline announcements included:

• Brand evolution: Snowflake Intelligence → Snowflake CoWork, Cortex Code → Snowflake CoCo
• Snowflake CoCo elevated to a full autonomous development platform with Desktop app, SDK, Cloud Agents, and integration with Vercel, Retool, and Superblocks
• Snowflake CoWork expanded with Deep Research, Artifacts (dashboards), User Memory, User Skills, Skill Catalog, and MCP connectors for Google Drive, Salesforce, and Slack
• Cortex Sense — a runtime context layer delivering 86% accuracy vs. 24% for generic models
• Snowflake Datastream — a fully managed Apache Kafka-compatible streaming service (private preview)
• Apache Iceberg V3 general availability with full bidirectional interoperability
• Open Semantic Interchange (OSI) — open standard with 54 participating vendors
• Horizon Context — three-part (Collect/Enrich/Activate) governed semantic foundation
• Cortex Training — custom model training on managed GPUs; 2× more training runs per GPU budget
• Agent Identity (GA) — cryptographic, verifiable identity for every AI agent
• AI Security Posture Management, data exfiltration policies, Cortex Guard
• Cortex AISQL and Dynamic Tables reaching General Availability
• $6 billion AWS partnership, Natoma acquisition (MCP gateway), Anthropic expanded partnership
• SAP Business Data Cloud Connect — bidirectional zero-copy integration
• dbt Labs + Fivetran merger announced at Summit
• Sanofi and Thomson Reuters featured as anchor enterprise customers

The company raised its FY27 growth guidance from 27% to 31% year-over-year, driven by unprecedented adoption of CoCo (7,100+ accounts) and CoWork (5,000+ accounts).

---

Event Overview & Theme

Item	Details
Event Name	Snowflake Summit 26
Dates	June 1–4, 2026
Location	Moscone Center, San Francisco, CA
In-Person Attendees	20,000+
Breakout Sessions	500+
Partners On-Site	200+
Hands-On Labs	39
Customer Speakers	300+
Session Tracks	12
Total New Capabilities	26+

Theme: The Agentic Enterprise

CEO Sridhar Ramaswamy opened the conference articulating his four-block vision for the agentic enterprise:

"The model is not your unique advantage. It's when you combine models with your data that things begin to shine."

Four Building Blocks (from the keynote):

1. Enterprise Data and Context — governed, AI-ready data foundation
2. AI Models — any model, without lock-in
3. Applications — CoWork for knowledge workers, CoCo for builders
4. Agentic Control Plane — the coordinating layer that makes agents enterprise-ready

Keynote Structure

Keynote	Date/Time	Hosts
Opening Keynote	June 1, 5:00 PM PDT	Sridhar Ramaswamy + Daniela Amodei (Anthropic President)
Platform Keynote	June 2, 9:00 AM PDT	Benoit Dageville + Christian Kleinerman (EVP Product)
Builder Keynote	June 3, 9:00 AM PDT	Live demos, Dev Day, Builder Theater

Platform Keynote YouTube recording: https://www.youtube.com/watch?v=NFTdHYp9uJo

---

Complete Capabilities Inventory

Snowflake announced 26+ new capabilities across six domains:

Capability	Domain	Status
Snowflake CoWork (Snowflake Intelligence rebrand)	AI Agents + Productivity	GA
Snowflake CoCo (Cortex Code rebrand)	AI Agents + Productivity	GA
CoWork Slackbot	AI Agents + Productivity	Private Preview soon
CoWork iOS App	AI Agents + Productivity	GA soon
CoWork Excel Extension	AI Agents + Productivity	Private Preview soon
Skill Catalog	AI Agents + Productivity	GA
User Skills	AI Agents + Productivity	Public Preview soon
Deep Research	AI Agents + Productivity	GA soon
Artifacts (publishable dashboards)	AI Agents + Productivity	GA soon
Snowflake Horizon Context	Context + Semantics	Private Preview (Wave 1)
Semantic Studio	Context + Semantics	Private Preview
Semantic View Autopilot	Context + Semantics	Public Preview
Cortex Sense	Context + Semantics	GA (in CoWork + CoCo)
Open Semantic Interchange (OSI)	Context + Semantics	Spec finalized
AI Agent Identity	Security + Governance	GA
AI Security Posture Management	Security + Governance	Public Preview
Data Exfiltration Policies	Security + Governance	Public Preview
Multi-Party Authorization	Security + Governance	Public Preview
Cortex Guard (prompt injection ML detection)	Security + Governance	Public Preview
Connected Audit Access	Security + Governance	Private Preview
External Engine Access Management	Security + Governance	Private Preview
Apache Iceberg V3 (read)	Infrastructure + Interoperability	GA (May 2026)
Snowflake Storage for Iceberg Tables	Infrastructure + Interoperability	GA
Zero-Copy Integrations (SAP, Salesforce, Workday, AVEVA, IBM)	Infrastructure + Interoperability	Public Preview
Snowflake Datastream	Infrastructure + Interoperability	Private Preview
Snowflake Openflow	Infrastructure + Interoperability	GA
Cortex Training	AI Infrastructure	Public Preview soon
Adaptive Compute	AI Infrastructure	GA soon
SpaceXAI Grok models	AI Infrastructure	GA
Auto-gen Agents for Data Shares	AI Infrastructure	Public Preview
Natoma acquisition (MCP gateway)	Partnerships + Ecosystem	Acquired
Snowflake + Anthropic expanded partnership	Partnerships + Ecosystem	GA
Snowflake + AWS $6B commitment	Partnerships + Ecosystem	Announced
Collibra bidirectional integration	Partnerships + Ecosystem	Preview (GA Q3 2026)

---

Major Rebrands: CoWork & CoCo

Snowflake Intelligence → Snowflake CoWork

The business-user-focused personal AI work agent has been officially renamed Snowflake CoWork.

• Target audience: Non-technical knowledge workers
• Rationale: Emphasizes the "working together" mission of enterprise AI
• Product page: https://www.snowflake.com/en/product/snowflake-cowork/

Cortex Code → Snowflake CoCo

Snowflake's AI coding agent has been renamed Snowflake CoCo.

• Target audience: Developers, data engineers, technical builders
• Rationale: Customers organically called it "CoCo" since launch in November 2025
• Product page: https://www.snowflake.com/en/product/features/cortex-code/
• Press release: https://www.snowflake.com/en/news/press-releases/snowflake-coco-redefines-enterprise-ai-development-as-the-coding-agent-built-for-faster-easier-and-more-powerful-innovation-anywhere

Adoption Curves

Product	Q3 FY26	Q4 FY26	Q1 FY27
CoWork	~1,200	2,500+	5,000+
CoCo (launched Nov 2025)	—	4,400	7,100+

Both represent "the fastest adoption of any new products in Snowflake's history."

---

Snowflake CoCo (formerly Cortex Code)

Overview

CoCo is Snowflake's AI coding agent purpose-built for enterprise data workflows. At Summit 2026, it evolved from an AI coding assistant into a full autonomous development platform.

Performance Benchmarks

On ADE-Bench (dbt Labs' framework for evaluating AI agents on real-world data engineering tasks):

Agent	Pass Rate
Snowflake CoCo	72.1%
Anthropic Claude Code	65.1%
OpenAI Codex	65.1%

CoCo also uses 51% fewer tokens and takes 8% less time than Claude Code with Opus 4.7. Two design principles drive efficiency:

1. Targeted exploration — navigates directly to relevant data rather than exhaustive scanning
2. Native tools approach — uses Snowflake, dbt, and Airflow tools natively instead of bash-based workarounds

New Features at Summit 2026

CoCo Desktop (GA Soon)

A native desktop application for Windows and macOS:

• One place to create data pipelines, build applications, design agents, debug notebooks, visualize data flows
• Automations: Schedule recurring tasks (pipeline refreshes, data quality checks, model retraining) governed by Snowflake RBAC with full audit trails
• Always-on persistent agent that maintains project context across sessions
• MCP integrations and Skill Catalog for reusable institutional best practices

Docs: https://docs.snowflake.com/en/LIMITEDACCESS/cortex-code-desktop

Cloud Agents

Serverless, event-driven workers running inside Snowsight:

• Each session gets a dedicated isolated container
• Execute shell commands, run Python scripts, install packages, read/write files
• Perform dbt builds and tests with dynamically generated Snowflake profile
• Search the web for additional context
• No local setup or infrastructure required

Docs: https://docs.snowflake.com/en/user-guide/cortex-code/cortex-code-snowsight/cloud-agents

VS Code Extension

CoCo as a first-class VS Code extension with inline schema browsing, query execution, and pipeline debugging.

Claude Code Plugin

Native integration with Anthropic Claude Code via Model Context Protocol:

• Plugin available at https://claude.com/plugins/snowflake-cortex-code
• Builders using Claude Code can invoke CoCo with full Snowflake schema/governance awareness

CoCo Agent SDK

Programmatic SDK for building custom applications on CoCo's agentic engine:

import { query } from "cortex-code-agent-sdk";

for await (const message of query({
  prompt: `Profile the ORDERS table in MY_DATABASE.ANALYTICS:
    - Total row count and date range covered
    - Null rate for each column
    - Top 5 customers by order volume
    Summarize findings in plain English.`,
  options: {
    cwd: process.cwd(),
    connection: "my-connection",
  },
})) {
  if (message.type === "assistant") {
    for (const block of message.content) {
      if (block.type === "text") process.stdout.write(block.text);
    }
  }
}

SDK features: multi-turn sessions, structured output (JSON schema validation), MCP server integration, hooks for agent behavior, streaming output, system prompts.

Docs: https://docs.snowflake.com/en/user-guide/cortex-code-agent-sdk/cortex-code-agent-sdk

MCP Server & ACP Support

• MCP server — CoCo is composable with any MCP-compatible orchestration layer
• ACP (Agent Client Protocol) — interoperability with 30+ dev environments (VS Code, Zed, JetBrains, Emacs)

Docs: https://docs.snowflake.com/en/user-guide/cortex-code/cortex-code-mcp

Docs: https://docs.snowflake.com/en/user-guide/cortex-code/cortex-code-acp

Application Integrations (New at Summit)

CoCo now integrates with popular application platforms:

• Vercel — deploy data-driven applications from CoCo directly
• Retool — build internal tools backed by CoCo's governed data queries
• Superblocks — enterprise app development with CoCo intelligence

Microsoft Excel Extension

CoCo as a Microsoft Excel extension — AI-assisted data analysis in spreadsheets.

Slack Integration (Coming Soon — Private Preview)

CoCo Slackbot: data workers can ask questions, trigger workflows, and share results in Slack. Answers are grounded in production data context and user access policies.

Mobile App (Coming Soon)

CoCo mobile app for iOS and Android — monitor pipelines, review task outputs, approve AI-generated workflows from anywhere.

Governance Built In

• All operations under user's existing Snowflake RBAC
• LLM inference stays within Snowflake's security perimeter
• Layered guardrails against prompt injection
• Prompt and response logging, query tagging, usage monitoring
• Admin-level cost and configuration controls

Resources

• Blog: https://www.snowflake.com/en/blog/snowflake-coco-ai-coding-agent-modern-data-stack
• Product: https://www.snowflake.com/en/product/features/cortex-code/

---

Snowflake CoWork (formerly Snowflake Intelligence)

Overview

Snowflake CoWork is the personal AI work agent for knowledge workers — now GA for all Snowflake customers. It is designed to move beyond question-and-answer to actually getting work done: "the system of action for the modern enterprise."

Press release: https://www.snowflake.com/en/news/press-releases/snowflake-cowork-powers-the-agentic-enterprise-as-the-personal-agent-for-knowledge-workers-to-work-smarter

New Features at Summit 2026

Deep Research (GA Soon)

Multi-step reasoning and analysis across structured AND unstructured enterprise data:

• Comprehensively searches all company data using multi-step reasoning
• Operates within the organization's governed data environment
• Delivers fully cited, actionable insights explaining the "why" behind numbers
• Powered by an agent swarm orchestration system built by Snowflake's AI Research Team
• Outperforms single-agent systems by over one-third on Snowflake's Hybrid Deep Research Benchmark
• Blog: https://www.snowflake.com/en/blog/engineering/arcticswarm-hybrid-deep-research

Artifacts (GA Soon)

Publishable interactive dashboards with:

• Governed views of live, trusted Snowflake data
• Knowledge workers explore via conversation (natural language), not just clicks
• Analysts create and publish dashboards; knowledge workers interact in CoWork
• Save, share, and collaborate — turning insights into reusable knowledge in daily workflows
• Governance and access policies built in from the start

User Memory (Public Preview Soon)

CoWork continuously learns from individuals' behavior:

• Adapts recommendations to the user's role over time
• Users can schedule tasks and receive timely, role-specific actions
• Works through Gmail and Slack integrations

User Skills (Public Preview Soon)

Automate unique workflows tailored to each user's day-to-day tasks:

• Turns everyday tasks into automated workflows
• Integrates with the Skill Catalog for enterprise-wide sharing and reuse

Skill Catalog (GA)

Enterprise-wide discovery and reuse of proven CoWork and CoCo skills:

• Teams share proven agent workflows across the organization
• Scales productivity gains beyond individual users

MCP Connectors (GA)

Native Model Context Protocol connectors for external enterprise systems:

• Google Drive — access and query Drive documents directly
• Salesforce — query CRM data, accounts, opportunities in CoWork
• Slack — bring Slack context into agent workflows
• Gmail — schedule tasks based on email context
• Built on Snowflake's Natoma acquisition for enterprise governance

Cortex Sense (GA in CoWork + CoCo)

See Section 7.

Personal Work Agent

Each user gets their own personalized agent with:

• Their own set of skills
• Their own MCP connectors
• Persistent memory and state across sessions
• Multi-agent orchestration — CoWork breaks complex requests into specialized sub-agents automatically

iOS Mobile App (GA Soon)

Native iOS interface for CoWork — knowledge workers take action on data from anywhere.

Slackbot (Private Preview Soon)

CoWork accessible directly in Slack for in-workflow data queries and governed dashboards.

Excel Extension (Private Preview Soon)

CoWork in Microsoft Excel for in-spreadsheet data queries.

Availability Status Summary

Feature	Availability
CoWork (core)	GA
Deep Research	GA soon
Artifacts / Publishable Dashboards	GA soon
User Memory	Public Preview soon
User Skills	Public Preview soon
Skill Catalog	GA
MCP connectors (Google Drive, Salesforce, Slack)	GA
iOS App	GA soon
Slackbot	Private Preview soon
Excel Extension	Private Preview soon
Cortex Sense	GA (within CoWork/CoCo)

Customer Outcomes

Company	Achievement
Synopsys	20+ purpose-built agents across revenue ops, legal, finance, product, IT — leaders query in natural language from Teams
WHOOP	Analyst-gated data now available to hundreds of employees in real time; reduced operational friction
Under Armour	Featured in Platform Keynote
TS Imagine	AI agent = 8.5 FTE; automated customer case resolution, trading/risk analysis
Toyota Motor Europe	Deployment timelines: months → weeks
Samsung	Real-time Galaxy S26 launch monitoring across all regions
United Rentals	5,000+ frontline employees query company data in natural language
Wolfspeed	Equipment troubleshooting: 2 hours → 2 minutes
Cisco	Large-scale data integration and decision acceleration

Snowflake Internal Productivity (Using CoWork)

Metric	Improvement
Job descriptions	60–120 min → 5–15 min
Employee survey analysis	Weeks → minutes
Customer project completion	Up to 5x faster
Response accuracy	+25%
Implementation cycle	Days → hours
Project margins	+40–50%
Customer go-live speed	+40% faster

---

Cortex Sense — Shared Context Layer

Overview

Cortex Sense is a runtime layer that automatically assembles data, business definitions, and operational knowledge that AI agents need — delivering strong out-of-the-box performance without manual context setup.

Performance Benchmark

Approach	Accuracy on Structured Questions
Cortex Sense (full business context)	86%
Generic frontier models alone	24%

What Cortex Sense Delivers

Component	Description
Business definitions	Centralized metric and dimension definitions
Operational knowledge	Business processes, workflows, and rules
Enterprise context	Org structure, ownership, team responsibilities
Prebuilt plugins	Role-specific plugins for finance, sales, etc.
MCP connectors	Standardized connections to external data sources

Cortex Sense works for both CoWork and CoCo, providing a unified context layer that agents use immediately.

Prebuilt Plugins

Cortex Sense ships with role-specific prebuilt plugins that combine:

• Skills and business logic
• MCP connectors
• Operational knowledge
• Production-ready from day one

Teams can go from concept to production in days instead of weeks.

---

Snowflake Datastream

Overview

Snowflake Datastream is a brand-new, fully managed Apache Kafka-compatible streaming service natively integrated into the Snowflake platform. Currently in private preview, it targets a $128 billion total addressable market.

Architecture

Component	Detail
Protocol	Full Apache Kafka API compatibility
Infrastructure	Fully managed — Snowflake handles brokers, scaling
Data landing	Governed Snowflake or Iceberg tables
BYOC option	Data plane in customer VPC available
Governance	RBAC, encryption, audit logging from day one
Analytics integration	Streaming and analytics side-by-side

Migration Path

Existing Kafka producers and consumers connect with a configuration change only — no code rewrites. The system eliminates the need for separate Kafka brokers, Confluent, Amazon MSK, or Azure Event Hubs clusters.

Use Case with AI

Datastream feeds real-time data directly to CoWork and CoCo agents, enabling:

• AI agents that respond to live streaming data
• Real-time anomaly detection in governed pipelines
• Immediate materialization of streaming events into queryable Snowflake tables

Resources

• Product page: https://www.snowflake.com/en/product/features/datastream
• Blog: https://www.snowflake.com/en/blog/data-development-simple-as-prompt
• Demo video: https://www.youtube.com/watch?v=68DP4ACQSjE

---

Apache Iceberg V3 & Open Interoperability

Overview

Snowflake reached GA for Apache Iceberg V3 and announced the broadest open interoperability framework in the industry, backed by Snowflake Horizon Catalog powered by Apache Polaris.

Press release: https://www.snowflake.com/en/news/press-releases/snowflake-pioneers-new-open-framework-for-interoperable-enterprise-data-and-ai

What's New in Iceberg V3

Feature	Status
Deletion vectors	GA
Row lineage	GA
Variant type (semi-structured data)	GA
Default column values	GA
External engines READ Snowflake Iceberg	GA
External engines WRITE to Snowflake Iceberg	Public Preview
Snowflake Storage for Iceberg Tables	GA
Iceberg REST Scan Plan API	Available

Bidirectional Interoperability

Via Snowflake Horizon Catalog (embedding open-source Apache Polaris):

• Apache Spark, Trino, Flink, Presto, Databricks, Microsoft Fabric — all read/write
• Microsoft Fabric: Real-world bidirectional testing validated in production
• Databricks Unity Catalog: Compatible

Zero-Copy Integrations (Public Preview)

No data duplication with:

• SAP Business Data Cloud
• Salesforce
• Workday
• AVEVA (industrial)
• IBM

Snowflake Storage for Iceberg Tables

New managed tier for open-format data:

• Single live governed copy of data across Snowflake and external lakes
• No data movement required
• Horizon Catalog governance on open-format tables

Snowflake's Commitment

Snowflake is on the Apache Polaris steering committee and contributes actively to ensuring Iceberg interoperability is real — not just marketing.

Resources

• Developer guide: https://www.snowflake.com/en/developers/guides/interoperable
• Engineering blog: https://www.snowflake.com/en/blog/engineering/bidirectional-interoperability-iceberg-snowflake-horizon-catalog
• Horizon vs Unity Catalog: https://www.snowflake.com/en/blog/engineering/iceberg-catalog-snowflake-horizon-vs-unity-catalog

---

Open Semantic Interchange (OSI)

Overview

Open Semantic Interchange (OSI) is a new open standard for sharing semantic models across data tools without vendor lock-in. The specification was finalized and announced at Summit 2026.

Participating Vendors (54 total)

Category	Companies
Governance/Catalog	Atlan, Collibra
Transformation	dbt Labs
Data Platforms	Snowflake, Databricks
Analytics	ThoughtSpot
Integration	Informatica
Financial Services	BlackRock
And 47 more	Full list at https://open-semantic-interchange.org/

What OSI Enables

• Business definitions (metrics, dimensions, KPIs) flow between tools without rewriting
• Semantic views defined in Snowflake are accessible to external AI agents and BI tools
• Avoids the semantic layer vendor lock-in problem
• Snowflake Horizon Context supports OSI, making trusted business definitions "universally accessible"

Strategic Significance

OSI signals that the industry has accepted semantic interoperability requires open formats — the same pattern that Iceberg validated for data storage. Snowflake's backing of OSI is an explicit bet that owning the semantic layer standard matters more than owning semantic layer lock-in.

---

Snowflake Horizon Context & Catalog Enhancements

Overview

Horizon Context is a new capability inside Snowflake Horizon Catalog that delivers a governed semantic foundation — ensuring every person, tool, and AI agent operates from the same version of the truth.

Press release: https://www.snowflake.com/en/news/press-releases/snowflake-advances-trusted-ai-with-snowflake-horizon-catalog-centralizing-governance-context-and-security-across-the-enterprise

Blog: https://www.snowflake.com/en/blog/horizon-context-governed-context

The Problem Horizon Context Solves

"An AI agent might recommend increasing prices based on revenue data, but if revenue is defined or calculated in different ways across systems, that recommendation can lead to misinformed decisions."

Traditional semantic layers are separate from the data itself. Horizon Context puts context directly alongside the data where it is governed and maintained.

Three-Part Architecture: Collect → Enrich → Activate

1. Collect — Trusted Business Context

Gather business logic fragmented across SQL, BI dashboards, and agents:

Wave 1 Connectors (Private Preview)	Description
PostgreSQL	Extract schema semantics from Postgres databases
SQL Server	Extract semantics from Microsoft SQL Server
Tableau	Import business logic from Tableau workbooks
Power BI	Import measure definitions from Power BI models
dbt	Extract semantic models and metric definitions from dbt projects

Additional connectors in future waves; OSI provides the standard interoperability layer.

2. Enrich — Automatically Maintain Business Context

Feature	Status	Description
Column-level lineage	GA	Stitch lineage across systems automatically
AI-generated documentation	Available	Auto-generate column/table descriptions
Popularity metrics	Available	Track which tables/columns are most queried
Semantic Studio	Private Preview	Define shared business logic without SQL
Semantic View Autopilot	Public Preview	Auto-creates and refines semantic views over time

Semantic View Autopilot also now auto-creates semantic views for data shared across organizations via Snowflake Marketplace, carrying trusted business context into every AI and analytics workflow.

3. Activate — Extend Trusted Context Everywhere

• CoCo integration: CoCo pulls rich business context from Horizon Context automatically, delivering more trustworthy responses even for data outside Snowflake
• External AI agent access: External agents can query governed business definitions via Horizon Context
• BI tool integration: Tableau, Power BI, and other tools access consistent metric definitions
• OSI compatibility: Business definitions portable to any OSI-supporting tool

Customer: BlackRock

"Snowflake's Horizon Context helps extend consistent business definitions across our broader data ecosystem, supporting more trusted and governed AI and analytics experiences at scale."
— Jeff Miller, Managing Director, Global Head of Data Factory & Enterprise Data Platforms, BlackRock

Adaptive Compute Integration

Working seamlessly with Horizon Catalog, Adaptive Compute automatically determines the optimal mix of compute and software resources in real time for AI and app workloads — without manual tuning or infrastructure management.

---

Cortex Training — Custom Model Training

Overview

Cortex Training extends Snowflake's existing Cortex Fine-tuning service into full custom model training on managed GPU infrastructure — entirely within Snowflake's governed environment.

Press release: https://www.snowflake.com/en/news/press-releases/snowflake-cowork-powers-the-agentic-enterprise-as-the-personal-agent-for-knowledge-workers-to-work-smarter

Key Capabilities

Capability	Detail
Training type	Fine-tuning + full custom training + reinforcement learning
Infrastructure	Fully managed GPU clusters (Snowflake-sourced)
Data governance	Data never leaves Snowflake environment
Efficiency	2× more training runs per equivalent GPU budget vs. alternatives
Supported models	Qwen family, Mistral family, others (open-weight)
Cost model	Pay-as-you-go, no cluster commitment

Why 2× Efficiency Matters

Cortex Training achieves 2× more training runs per GPU budget through:

• Optimized scheduling of GPU capacity
• No cold-start overhead from external clusters
• Direct data access without movement overhead

Featured Customer: Resolve AI

Resolve AI (AI for running and operating software in production) made a multi-million-dollar, two-year commitment to Cortex Training to build domain-specific models:

• Uses reinforcement learning (RL) on proprietary training data
• Builds models on multi-agent harnesses, simulated environments, and evaluation workflows
• Cortex Training provides governed infrastructure to run RL training workloads continuously
• Models outperform general-purpose alternatives on production engineering reasoning tasks
• Snowflake itself is a Resolve AI customer — Snowflake engineering uses Resolve AI in production workflows

"General-purpose API models will continue to improve, but there is a limit to how far you can get when building on top of them. Production operations for running and managing complex software systems have specific failure modes, reasoning patterns, and the highest bar for accuracy."
— Spiros Xanthos, Founder and CEO, Resolve AI

Status

Cortex Training is in public preview soon. The companion Cortex Fine-tuning service is already GA.

Docs: https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-finetuning

---

AI Security for the Agentic Enterprise

Overview

Summit 2026 introduced a complete zero-trust security suite specifically for AI agents. The organizing frame: traditional security was built for human users, not autonomous AI agents.

Press release: https://www.snowflake.com/en/news/press-releases/snowflake-advances-trusted-ai-with-snowflake-horizon-catalog-centralizing-governance-context-and-security-across-the-enterprise

Blog: https://www.snowflake.com/en/blog/enterprise-ai-security/

McKinsey finding cited: "Nearly two-thirds of organizations cite security as the top barrier to scaling AI."

New Security Capabilities

Agent Identity (GA) ✅

Cryptographic, verifiable identity for every AI agent before data access:

• Per-agent RBAC — NOT inherited from user credentials
• Dynamic data masking by agent type
• Complete audit trail of every single agent activity
• Prevents rogue agent actions
• Prevents privilege escalation through agent chains

AI Security Posture Management (Public Preview)

Continuous monitoring within Snowflake Trust Center:

• Monitor security posture of AI systems continuously
• Investigate violations faster with AI-guided, context-aware assistance
• Accelerate risk response
• Reduce alert fatigue as AI workloads scale

Data Exfiltration Policies (Public Preview)

• Block specific data movements during agent operations
• Granular control per operation type
• Prevent unauthorized data exposure through agentic workflows

Multi-Party Authorization (Public Preview)

• Require two administrators or two users to approve sensitive agent actions
• Four-eyes principle for high-risk operations
• Full audit trail on all approvals

Cortex Guard (Public Preview)

ML-driven threat detection with:

• Prompt injection detection — block jailbreak attempts
• Detection of emerging zero-day vulnerabilities
• Stops threats at machine speed
• Works inline in agent workflows without slowing innovation

Connected Audit Access (Private Preview)

Unified audit access within Horizon Catalog for cross-system visibility — see exactly what every agent accessed, when, and why.

External Engine Access Management (Private Preview)

Controls access to Snowflake data from external compute engines (Spark, Databricks, etc.) — apply governance to data accessed from outside Snowflake's perimeter.

Security Customers

Enterprises exploring these new AI security capabilities:

• Acxiom — securing AI access to personally identifiable data
• NewDay — scaling AI with compliance accountability
• Thomson Reuters — securing AI access to sensitive professional data

---

Cortex AISQL — General Availability

Overview

Cortex AISQL reached General Availability at Summit 2026. Run AI inference directly inside SQL queries, unified across structured and unstructured data.

Core Functions

Function	Description
AI_FILTER	Filter rows using natural language criteria
AI_AGG	Aggregate unstructured content with AI synthesis
AI_EXTRACT	Extract structured fields from unstructured text
AI_CLASSIFY	Classify text or content into categories
AI_SENTIMENT	Analyze sentiment of text

Example

SELECT
    o.customer_id,
    o.order_value,
    AI_CLASSIFY(t.transcript_text, ['positive', 'negative', 'neutral']) AS sentiment
FROM orders o
JOIN support_transcripts t ON o.customer_id = t.customer_id
WHERE o.order_date >= '2026-01-01';

AI Redact (Public Preview Coming Soon)

PII and sensitive data detection inside AISQL pipelines:

• Auto-detect SSNs, account numbers, HIPAA-relevant identifiers
• Scrub fields before they reach AI models
• Removes a major procurement blocker for regulated industries

---

Dynamic Tables — General Availability

Overview

Dynamic Tables reached General Availability — enabling declarative, continuously refreshed AI inference pipelines.

Example: Continuous Sentiment Pipeline

CREATE OR REPLACE DYNAMIC TABLE customer_sentiment
    TARGET_LAG = '5 minutes'
    WAREHOUSE = analytics_wh
AS
SELECT
    review_id,
    customer_id,
    review_text,
    AI_SENTIMENT(review_text) AS sentiment_score,
    AI_CLASSIFY(review_text, ['positive', 'neutral', 'negative']) AS sentiment_label,
    CURRENT_TIMESTAMP() AS last_refreshed
FROM raw_customer_reviews;

• Write the AI SQL once
• Snowflake handles incremental refresh automatically
• No orchestration code required

Docs: https://docs.snowflake.com/en/user-guide/dynamic-tables-intro

---

Snowflake CoWork / Intelligence — General Availability

CoWork is now Generally Available to all Snowflake customers (12,000+).

Metric	Value
Customers with GA access	All 12,000+
Agents deployed in production (early access)	15,000+
Active accounts (Q1 FY27)	5,000+

Developer Mode is included in GA — build custom agents with Snowflake's governed data context and audit logging.

---

Adaptive Compute — Expanded Availability

Overview

Adaptive Compute (self-sizing warehouse) expanded regional availability and is GA soon.

How It Works

CREATE WAREHOUSE my_adaptive_wh
    WAREHOUSE_TYPE = ADAPTIVE
    MAX_QUERY_PERFORMANCE_LEVEL = 8;

• No manual sizing (XS–6XL) — compute is allocated dynamically per query
• Small queries → small allocations; large analytical queries → more compute
• Set MAX_QUERY_PERFORMANCE_LEVEL cap; Snowflake routes each query to optimal allocation
• Compatible with ACCOUNT_USAGE views, resource monitors, and budgets

Integration with Horizon Catalog

Adaptive Compute works alongside Horizon Catalog security and governance, providing a "true serverless experience" that combines governance with operational efficiency.

Blog: https://www.snowflake.com/en/blog/adaptive-compute-performance/

---

Openflow — Wider General Availability

Overview

Openflow — Snowflake's managed data integration service built on Apache NiFi — reached wider GA at Summit 2026.

Architecture

┌─────────────────────────────────────┐
│  Snowflake (Managed Control Plane)  │
│ - Orchestration, monitoring, config │
└──────────────────┬──────────────────┘
                   │ Control signals only
┌──────────────────▼──────────────────┐
│  Customer VPC (BYOC Data Plane)     │
│  - Apache NiFi runtime              │
│  - Actual data movement             │
│  - Your network, your rules         │
└─────────────────────────────────────┘

BYOC answers the compliance question: data never leaves the customer's environment en route to Snowflake.

Connectors

Connector	Status
Oracle DB	GA (February 2026)
Microsoft SQL Server	Available
PostgreSQL	Available
CDC sources	Available
Event streaming	Available

Product: https://www.snowflake.com/en/product/features/openflow/

---

SAP Business Data Cloud Partnership

Overview

Bidirectional, zero-copy data sharing between SAP Business Data Cloud and Snowflake — announced at Summit 2026.

Two Products

Product	Description
SAP Snowflake (solution extension)	Snowflake AI capabilities natively inside SAP BDC
SAP BDC Connect for Snowflake	Bidirectional zero-copy: SAP data in Snowflake and Snowflake data in SAP

Zero-Copy Means

• No data movement = no synchronization latency
• No data duplication = no storage cost doubling
• Both platforms see live data simultaneously
• No ETL jobs ever needed between SAP and Snowflake analytics

Use Case

Run Cortex AISQL against live SAP finance data. Build CoWork agents on live SAP supply chain data. Train custom models on SAP HR data via Cortex Training. All in real time. All without an ETL job.

Availability: GA targeted H1 2026.

---

AWS Partnership & $6B Commitment

Press release: https://www.snowflake.com/en/news/press-releases/snowflake-expands-aws-collaboration-with-6b-commitment-to-accelerate-enterprise-agentic-ai-adoption

Item	Detail
Commitment	$6 billion over 5 years
Infrastructure	AWS Graviton chips for Snowflake AI
AI Services	Deeper AWS AI service integration
Region expansion	10 new AWS regions
Purpose	Power the orchestration layer for Snowflake's agentic platform

---

Natoma Acquisition — MCP Gateway

Press release: https://www.snowflake.com/en/news/press-releases/snowflake-announces-intent-to-acquire-natoma-providing-secure-connectivity-for-the-agentic-enterprise

Item	Detail
Announcement	Intent to acquire announced May 27, confirmed at Summit
What Natoma provides	Enterprise MCP (Model Context Protocol) gateway
Capabilities	Centralized MCP gateway with identity verification, access policies, audit controls at tool-call level
Scope	Governs what actions AI agents can take against APIs, workflows, collaboration systems, and operational systems
Integration	Powers CoWork MCP connectors for Google Drive, Salesforce, Slack

Natoma governs what agents are allowed to do at the external tool level — complementing Agent Identity, which governs who the agent is within Snowflake.

---

Anthropic Partnership Expansion

Press release: https://www.snowflake.com/en/news/press-releases/snowflake-and-anthropic-accelerate-enterprise-ai-adoption-driven-by-rising-demand-for-governed-ai

Building on the December 2025 expanded partnership ($200M), Summit 2026 formalized the following:

Key Co-Innovation Areas

Area	Detail
Claude in Cortex AI	All Claude models (Haiku, Sonnet, Opus, Claude 4) directly in Cortex AI across all major cloud platforms
CoWork powered by Claude	Claude powers reasoning in Snowflake CoWork for natural language queries and insight synthesis
CoCo powered by Claude	Claude powers the coding intelligence in Snowflake CoCo
Claude Code plugin	CoCo available as plugin in Claude Code at https://claude.com/plugins/snowflake-cortex-code
Claude Marketplace	Snowflake is one of six launch partners in Claude Marketplace — simplifies procurement and unlocks joint commercial models
Claude Code Security	Joint development on security capabilities to identify, assess, and remediate vulnerabilities with built-in human oversight

Daniela Amodei at Summit

Anthropic President Daniela Amodei co-keynoted with Sridhar Ramaswamy, defining the enterprise AI moment:

"Snowflake brings the governed data environment enterprises already rely on, and Claude brings the reasoning to put that data to work. Together we're making it easy for organizations to use trusted AI on their most critical business data."
— Steve Corfield, Head of Global Business Development & Partnerships, Anthropic

Customer Momentum (Anthropic + Snowflake)

Customers using Claude in Snowflake Cortex AI: Basis, Block, Carvana, eSentire, Indeed, Notion, and more.

---

Developer Experience: Workspaces & Git

Snowflake Workspaces

Collaborative development environment inside Snowflake:

• Multiple team members build on the same data simultaneously
• Replaces coordination across multiple external tools
• Integrated with CoCo Desktop

Native Git Integration

• Version-controlled SQL, Python, and pipeline code directly from Snowflake
• Full commit, branch, and PR workflows from Snowsight
• Audit trail on all code changes

Snowflake Dev Day — June 25 (Virtual)

Post-Summit developer event:

• "Don't just hear about AI — build it"
• Luminary talks and hands-on labs
• Free registration: https://www.snowflake.com/en/events/

---

AI Models & Model Marketplace

Cortex AI Model Catalog

Provider	Models
Anthropic	Claude 3.5 Sonnet+, Haiku, Opus, Claude 4 series
OpenAI	GPT-4o, GPT-4o mini
Google	Gemini family
Meta	Llama 3.x family
Mistral AI	Mistral Large, 7B, Mixtral
DeepSeek	DeepSeek-V3, DeepSeek-R1
SpaceXAI	NEW at Summit: Grok models (GA)
Open-weight (trainable)	Qwen, Mistral, Llama via Cortex Training

Auto-gen Agents for Data Shares (Public Preview)

Snowflake Marketplace datasets now automatically generate:

• Cortex Agents ready to query the dataset
• Semantic Views carrying trusted business context
• Deployed with no manual configuration

This opens Snowflake Marketplace to AI-native data products.

Model Selection Philosophy

"Models keep changing, capabilities keep advancing. But the data is constant. Our offer to customers: we help AI models connect to enterprise data and enterprise context with security in mind."
— Christian Kleinerman, EVP Product, Snowflake

---

dbt Labs + Fivetran Merger

Announced at Summit 2026: dbt Labs and Fivetran are merging into a single company.

Item	Detail
Companies	dbt Labs (transformation) + Fivetran (data integration)
Combined vision	"Trusted agents using open data infrastructure principles"
dbt Core v2.0	In alpha; supports Snowflake, BigQuery, Databricks, Redshift
Relevance to Snowflake	Atlan, dbt, Fivetran, and Snowflake all participating in OSI

This merger creates a unified open data infrastructure stack and deepens Snowflake's partner ecosystem. dbt's semantic layer and Fivetran's connectors will likely align with OSI standards.

---

Collibra Integration

Announced at Summit 2026: Snowflake and Collibra announced an expanded bidirectional integration.

Item	Detail
Direction	Bidirectional metadata flow
From Collibra → Snowflake	Governed metadata, tags, and policies into Horizon Catalog and Cortex
From Snowflake → Collibra	Technical metadata back to Collibra
Semantic alignment	Semantic models aligned to OSI published as semantic views
Status	Preview (GA targeted Q3 2026)

Collibra-governed business glossary terms flow into Snowflake's AI context layer, enabling CoWork and CoCo to answer with enterprise-defined vocabulary.

---

Auto-gen Agents for Data Shares

Status: Public Preview

Snowflake Marketplace datasets now automatically generate AI agents and semantic views:

• New listings come pre-equipped with Cortex Agents
• Agents can answer natural language questions about the dataset immediately
• Semantic Views carry governed business context
• Reduces time-to-value for marketplace data consumers from days to minutes

---

Snowflake Postgres

Overview

Snowflake's managed PostgreSQL-compatible OLTP service continues to evolve as a key part of the platform strategy.

Capability	Status
Full PostgreSQL wire protocol compatibility	GA
OLTP workloads alongside Snowflake analytics	GA (Unistore architecture)
pg_lake → Iceberg tables integration	Available
Catalog integration via CATALOG_SOURCE = SNOWFLAKE_POSTGRES	Available

pg_lake + Iceberg

• Snowflake Postgres exposes tables as Iceberg tables via pg_lake
• These tables are readable by Snowflake Analytics using catalog integration
• Enables analytics on live OLTP data without ETL

---

Partner Ecosystem Highlights

Announcements by Partner

Partner	Announcement
Anthropic	Expanded partnership; Daniela Amodei keynote; Claude in Cortex + CoWork + CoCo; Claude Marketplace launch partner
SAP	BDC Connect zero-copy bidirectional integration
AWS	$6B 5-year commitment; Graviton infrastructure; 10 new regions
Microsoft	Fabric Iceberg bidirectional (validated); Excel extensions for CoCo/CoWork
Salesforce	Zero-copy integration; MCP connector via Natoma
Workday	Zero-copy integration; MCP connector via Natoma
Aveva	Zero-copy industrial data integration
IBM	Zero-copy data integration
Collibra	Bidirectional metadata flow; GA Q3 2026
Atlan	OSI founding member; context layer for Snowflake
Elementum	AI-native SaaS replacement; Sanofi deployment
Resolve AI	Multi-million dollar Cortex Training commitment
dbt Labs + Fivetran	Merger announced; OSI participation
Accenture	Featured in opening keynote; enterprise AI deployment
Deloitte	Enterprise AI deployment partner; Jason Salzetti quote

Snowflake Dev Day (Post-Summit)

Free virtual event June 25, 2026:

• Hands-on labs with the full agentic data stack
• Luminary talks from Snowflake engineers
• Build your first agent, semantic view, and CoCo workflow

Partner Awards

• Snowflake Partner Awards — recognized innovation and impact across the global ecosystem
• Data Drivers Awards — customer recognition for driving business impact

---

Customer Success Stories

Flagship: Thomson Reuters

Press release: https://www.snowflake.com/en/news/press-releases/thomson-reuters-powers-trusted-enterprise-ai-at-scale-on-snowflake

Metric	Value
Internal users	1,500+ daily
Tables governed	37,500+
Data sources connected	350
Workload performance	Up to 3.4× faster
Complex analysis speed	Weeks → seconds
Snowflake on platform since	2021

Thomson Reuters uses CoCo to modernize legacy systems. Key products powered: CoCounsel (AI for legal), Westlaw. AI security capabilities explored to scale governed enterprise AI responsibly.

---

Flagship: Sanofi

Press release: https://www.snowflake.com/en/news/press-releases/sanofi-chooses-snowflake-to-accelerate-its-ai-powered-drug-development

Customer video: https://www.youtube.com/watch?v=iu9up54Dl6Q

Goal: Become the first biopharma company powered by AI at scale.

Area	Achievement
"Concierge for Field"	AI agent that prepares sales reps for every physician visit — pre-call plan in seconds
Domains	R&D, procurement, IT, HR, field sales
Partner	Elementum (AI-native SaaS replacement) runs directly on Snowflake
Data	Clinical data at scale processed in Snowflake for R&D decisions
Key quote	"We are building AI directly on our data and reinventing how the company runs" — Emmanuel Frenehard, CDO, Sanofi

---

CoCo Customer Outcomes

Company	Achievement
Thomson Reuters	Insights in days vs. weeks; 37,500+ governed tables; 3.4× faster workloads
Fanatics	Dev timelines from weeks to hours; FanGraph democratized for business users
evolv	500+ hours saved in first 20 days; ~$100K in value generated
Infinite Lambda	Complete Customer 360 app built in 5 hours
WHOOP	Developer productivity doubled; 100+ workflows automated in weeks
Block	AI agents for compliance, security, trading across Square, Cash App, Afterpay

CoWork Customer Outcomes

Company	Achievement
Synopsys	20+ agents across 5 departments; leaders query in natural language from Teams
WHOOP	Analyst-gated data accessible to hundreds of employees in real time
Under Armour	Featured in Platform Keynote
TS Imagine	AI agent = 8.5 FTE; trading/risk management automation
Toyota Motor Europe	Deployment timelines: months → weeks
Samsung	Real-time Galaxy S26 launch visibility across regions
United Rentals	5,000+ frontline users with natural language data queries
Wolfspeed	Equipment troubleshooting: 2 hours → 2 minutes
Indeed	Self-service analytics across all employees; faster hiring platform improvements
Carvana	Dynamic inventory, logistics, financing AI — secure, governed, at scale
Notion	Data Scout agent pulls from Snowflake to surface real-time business insights
Basis	AI-driven marketing automation with transparency and governance

Snowflake Internal Results (Using Own Products)

Metric	Value
CoCo adoption	7,100+ customer accounts
Automated workflows created (internally)	100+ across finance, HR, marketing, sales
Case resolution improvement	25%
Engineer throughput increase	25%
Developer productivity	Doubled (PRs + code output)
Productivity equivalent recovered (sales org)	~90 FTE annually
CoCo user count (Snowflake internal)	Growing rapidly

---

Key Keynotes & Speakers

Opening Keynote — June 1, 2026

Speaker	Role	Key Message
Sridhar Ramaswamy	CEO, Snowflake	Agentic enterprise vision; four building blocks; data as the constant advantage
Daniela Amodei	Co-Founder & President, Anthropic	Enterprise AI requires grounded data; Snowflake + Anthropic = governed AI at scale
Accenture leader	Global SI partner	Enterprise AI deployment at scale
Sanofi CDO	Emmanuel Frenehard	First biopharma powered by AI at scale

Platform Keynote — June 2, 2026

Speaker	Role	Topic
Benoit Dageville	Co-Founder & President of Product	Open architecture vision; interoperability
Christian Kleinerman	EVP of Product	CoWork/CoCo rebrands; Datastream; Iceberg V3; Security; OSI
Caitlin Halferty	Head of Data & Analytics, Thomson Reuters	CoCo deployment at scale; 3.4× workload improvement
Under Armour	Customer	CoWork deployment

YouTube recording: https://www.youtube.com/watch?v=NFTdHYp9uJo

Builder Keynote — June 3, 2026

• Live demos of the full agentic data stack
• Dev Day with hands-on labs and community sessions
• Snowflake Startup Challenge Finale — winner announced from 100+ country applicants; up to $1M investment opportunity

Special Events

Event	Description
CTO Circle	Invite-only for CTOs scaling AI in production; engineering velocity and risk management
Dev Day	Builder-focused; Gen AI bootcamps; Startup Challenge
Basecamp Expo Hall	200+ partner demos; 1:1 expert consultations
Braindate	Peer-to-peer connection platform integrated into Summit app
Executive Program	Private boardroom discussions for senior leaders

---

Platform Metrics & Business Impact

Customer and Usage Metrics

Metric	Value
Total Snowflake customers	13,900+
Global 2000 customers	810
Customers using AI weekly	13,600
CoCo customer accounts	7,100+
CoWork customer accounts	5,000+
Agents deployed in production	15,000+
Summit attendees (in-person)	20,000+

Business Impact

Metric	Value
FY27 growth guidance (raised at Summit)	31% YoY (up from 27%)
CEO characterization	"Fastest adoption of any new products in company history"
Total new capabilities announced	26+

CoWork Adoption Curve

Q3 FY26  █████████████  ~1,200 accounts
Q4 FY26  ██████████████████████████  2,500+ accounts  (+108%)
Q1 FY27  █████████████████████████████████████████████  5,000+ accounts  (+100%)

CoCo Adoption Curve

Q4 FY26  ███████████████████████████████████████  4,400 accounts (launch quarter)
Q1 FY27  ██████████████████████████████████████████████████████  7,100+ accounts  (+61%)

---

Official Press Releases

All press releases published at Snowflake Summit 2026:

Title	Date	URL
Snowflake Makes AI Real for Businesses at Snowflake Summit 26	April 22 (pre-Summit)	https://www.snowflake.com/en/news/press-releases/snowflake-makes-ai-real-at-snowflake-summit-26-featuring-anthropics-daniela-amodei-and-other-industry-leaders/
Snowflake Announces Intent to Acquire Natoma	May 27	https://www.snowflake.com/en/news/press-releases/snowflake-announces-intent-to-acquire-natoma-providing-secure-connectivity-for-the-agentic-enterprise/
Snowflake Expands AWS Collaboration with $6B Commitment	May 27	https://www.snowflake.com/en/news/press-releases/snowflake-expands-aws-collaboration-with-6b-commitment-to-accelerate-enterprise-agentic-ai-adoption/
Snowflake Q1 FY27 Financial Results	May 27	https://investors.snowflake.com/news/news-details/2026/Snowflake-Reports-Financial-Results-for-the-First-Quarter-of-Fiscal-2027/default.aspx
Snowflake and Anthropic Accelerate Enterprise AI Adoption	June 1	https://www.snowflake.com/en/news/press-releases/snowflake-and-anthropic-accelerate-enterprise-ai-adoption-driven-by-rising-demand-for-governed-ai/
Thomson Reuters Powers Trusted Enterprise AI at Scale on Snowflake	June 2	https://www.snowflake.com/en/news/press-releases/thomson-reuters-powers-trusted-enterprise-ai-at-scale-on-snowflake/
Sanofi Chooses Snowflake to Accelerate AI-Powered Drug Development	June 2	https://www.snowflake.com/en/news/press-releases/sanofi-chooses-snowflake-to-accelerate-its-ai-powered-drug-development/
Snowflake CoWork Powers the Agentic Enterprise	June 2	https://www.snowflake.com/en/news/press-releases/snowflake-cowork-powers-the-agentic-enterprise-as-the-personal-agent-for-knowledge-workers-to-work-smarter/
Snowflake CoCo Redefines Enterprise AI Development	June 2	https://www.snowflake.com/en/news/press-releases/snowflake-coco-redefines-enterprise-ai-development-as-the-coding-agent-built-for-faster-easier-and-more-powerful-innovation-anywhere/
Snowflake Advances Trusted AI with Snowflake Horizon Catalog	June 2	https://www.snowflake.com/en/news/press-releases/snowflake-advances-trusted-ai-with-snowflake-horizon-catalog-centralizing-governance-context-and-security-across-the-enterprise/
Snowflake Pioneers New Open Framework for Interoperable Enterprise Data and AI	June 2	https://www.snowflake.com/en/news/press-releases/snowflake-pioneers-new-open-framework-for-interoperable-enterprise-data-and-ai/

---

Resources, Links & Repositories

Official Snowflake Resources

Resource	URL
Summit 26 Homepage	https://www.snowflake.com/en/summit/
Summit On-Demand Sessions	https://events.snowflake.com/summit/on-demand/agenda/
Snowflake Newsroom	https://www.snowflake.com/en/news/
Snowflake Blog	https://www.snowflake.com/en/blog/
Snowflake Documentation	https://docs.snowflake.com/
Developer Guides	https://www.snowflake.com/en/developers/guides

Product Documentation

Product	URL
Snowflake CoCo	https://docs.snowflake.com/en/user-guide/cortex-code/
CoCo Desktop	https://docs.snowflake.com/en/LIMITEDACCESS/cortex-code-desktop
CoCo Agent SDK	https://docs.snowflake.com/en/user-guide/cortex-code-agent-sdk/cortex-code-agent-sdk
CoCo Cloud Agents	https://docs.snowflake.com/en/user-guide/cortex-code/cortex-code-snowsight/cloud-agents
CoCo MCP Server	https://docs.snowflake.com/en/user-guide/cortex-code/cortex-code-mcp
CoCo ACP	https://docs.snowflake.com/en/user-guide/cortex-code/cortex-code-acp
Cortex Fine-tuning	https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-finetuning
Dynamic Tables	https://docs.snowflake.com/en/user-guide/dynamic-tables-intro
Adaptive Warehouses	https://docs.snowflake.com/en/user-guide/warehouses-adaptive
Snowflake CoWork	https://www.snowflake.com/en/product/snowflake-cowork/
Datastream	https://www.snowflake.com/en/product/features/datastream
Cortex Agents	https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents

Blog Posts

Topic	URL
CoCo Summit blog	https://www.snowflake.com/en/blog/snowflake-coco-ai-coding-agent-modern-data-stack
CoWork Summit blog	https://www.snowflake.com/en/blog/snowflake-cowork-personal-work-agent
Cortex AI enterprise platform	https://www.snowflake.com/en/blog/snowflake-for-ai-enterprise-ai-platform
Horizon Context blog	https://www.snowflake.com/en/blog/horizon-context-governed-context
AI Security blog	https://www.snowflake.com/en/blog/enterprise-ai-security/
Adaptive Compute blog	https://www.snowflake.com/en/blog/adaptive-compute-performance/
Trusted Data/AI blog	https://www.snowflake.com/en/blog/trusted-data-trusted-ai
Agentic control plane	https://www.snowflake.com/en/blog/agentic-enterprise-control-plane/
Data Development as Prompt	https://www.snowflake.com/en/blog/data-development-simple-as-prompt
Iceberg Bidirectional	https://www.snowflake.com/en/blog/engineering/bidirectional-interoperability-iceberg-snowflake-horizon-catalog
ArcticSwarm Deep Research	https://www.snowflake.com/en/blog/engineering/arcticswarm-hybrid-deep-research
Semantic View Autopilot	https://www.snowflake.com/en/blog/semantic-view-autopilot/

Open Source Projects

Project	Purpose	URL
Apache Polaris	Open-source Iceberg catalog (Snowflake donated)	https://github.com/apache/polaris
ADE-Bench	dbt Labs benchmark for AI coding agents	https://github.com/dbt-labs/ade-bench
Apache Iceberg	Open table format	https://github.com/apache/iceberg
Apache NiFi	Data integration framework (powers Openflow)	https://github.com/apache/nifi
Model Context Protocol	Open standard for AI tool connectivity	https://github.com/modelcontextprotocol/specification
Open Semantic Interchange	Open standard for semantic model sharing	https://open-semantic-interchange.org/

Media Coverage

Outlet	Article
SiliconAngle	https://siliconangle.com/2026/06/02/ai-agents-open-data-governance-take-center-stage-snowflake-summit/
Constellation Research	https://www.constellationr.com/insights/news/snowflake-summit-2026-context-custom-model-training-iceberg-v3
Diginomica	https://diginomica.com/snowflake-summit-2026-how-snowflake-making-strategic-shift-towards-agentic-ai
Atlan	https://atlan.com/know/snowflake/summit-2026-announcements/
theCUBE Coverage	https://www.thecube.net/events/snowflake/snowflake-summit-2026
ChatForest Recap	https://chatforest.com/builders-log/snowflake-summit-26-recap-intelligence-ga-cortex-code-openflow-agentic-data-stack/
Row Zero Insights	https://rowzero.com/blog/insights-from-snowflake-summit-2026
Futurum Group	https://futurumgroup.com/insights/snowflake-summit-2026-four-infrastructure-bets-that-determine-whether-the-agentic-enterprise-delivers/

Videos

Video	URL
Platform Keynote	https://www.youtube.com/watch?v=NFTdHYp9uJo
Snowflake Datastream Demo	https://www.youtube.com/watch?v=68DP4ACQSjE
Snowflake AI Pulse: Latest Features	https://www.youtube.com/watch?v=ELn8eMi3tQk
Cortex Fine-tuning for Enterprise	https://www.youtube.com/watch?v=vq9zYypx7_c
Sanofi Customer Video	https://www.youtube.com/watch?v=iu9up54Dl6Q
Thomson Reuters Customer Video	https://www.snowflake.com/en/customers/all-customers/video/thomson-reuters/

Developer Resources (Getting Started)

Resource	URL
Build agentic application quickstart	https://www.snowflake.com/en/developers/guides/build-agentic-application-in-snowflake/
Cortex Agents docs	https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents
Claude Code Plugin	https://claude.com/plugins/snowflake-cortex-code
Dev Day (June 25)	https://www.snowflake.com/en/events/

Internal Snowflake Resources

Resource	Access
PLT Data	SALES.RAVEN.PLT_DATA_VIEW via Snowhouse
Summit On-Demand Recordings	https://events.snowflake.com/summit/on-demand/agenda/
Internal Slack	#summit-26, #coco-feedback, #cowork-internal
What's New Sessions	Available on Summit on-demand portal

---

Feature Availability Summary

Feature	Status at Summit 2026
Snowflake CoWork (core)	Generally Available
CoWork Deep Research	GA Soon
CoWork Artifacts (dashboards)	GA Soon
CoWork iOS Mobile App	GA Soon
CoWork User Memory	Public Preview Soon
CoWork User Skills	Public Preview Soon
CoWork Skill Catalog	Generally Available
CoWork MCP Connectors (Google Drive, Salesforce, Slack)	Generally Available
CoWork Slackbot	Private Preview Soon
CoWork Excel Extension	Private Preview Soon
Snowflake CoCo (core)	Generally Available
CoCo Desktop	GA Soon
CoCo Cloud Agents	Generally Available
CoCo VS Code Extension	Generally Available
CoCo Agent SDK	Generally Available
CoCo MCP Server	Generally Available
CoCo ACP Support	Generally Available
CoCo Excel Extension	Available
CoCo Slack Integration	Private Preview (Coming Soon)
CoCo Mobile App	Coming Soon
CoCo Vercel / Retool / Superblocks integrations	Available
Cortex Sense	Generally Available (in CoWork + CoCo)
Cortex Training	Public Preview Soon
Cortex AISQL	Generally Available
AI Redact (in AISQL)	Public Preview Coming Soon
Dynamic Tables	Generally Available
Adaptive Compute	GA Soon
Apache Iceberg V3 (read access by external engines)	Generally Available
Apache Iceberg V3 (write access by external engines)	Public Preview
Snowflake Storage for Iceberg Tables	Generally Available
Iceberg REST Scan Plan API	Available
Horizon Context (Collect/Enrich/Activate)	Private Preview (Wave 1)
Semantic Studio	Private Preview
Semantic View Autopilot	Public Preview
Connected Audit Access	Private Preview
External Engine Access Management	Private Preview
Open Semantic Interchange (OSI)	Spec Finalized
Agent Identity	Generally Available
AI Security Posture Management	Public Preview
Data Exfiltration Policies	Public Preview
Multi-Party Authorization	Public Preview
Cortex Guard (prompt injection detection)	Public Preview
Snowflake Datastream	Private Preview
Openflow (wider GA)	Generally Available
SAP BDC Connect	GA Targeted H1 2026
Zero-Copy Integrations (SAP, Salesforce, Workday, AVEVA, IBM)	Public Preview
SpaceXAI Grok models	Generally Available
Auto-gen Agents for Data Shares	Public Preview
Collibra bidirectional integration	Preview (GA Q3 2026)
Natoma MCP Gateway	Available (Acquired)

---

Summary: The Agentic Enterprise Vision

Snowflake's Four Strategic Pillars at Summit 2026

Pillar	What It Means	Key Announcements
AI-Ready Data	Governed, clean, semantically rich data	Iceberg V3, Horizon Context, Dynamic Tables GA, AI Redact
Model Flexibility	Any model without lock-in	SpaceXAI Grok, Cortex Training, all frontier models
Interoperability	Open, bidirectional connectivity	Iceberg V3 bidirectional, OSI (54 vendors), SAP/Salesforce/Workday/Fabric zero-copy
Agentic Control Plane	Operating system for enterprise intelligence	CoCo Desktop/SDK/Cloud Agents, CoWork Deep Research/Artifacts/Skills

The Agentic Stack (Bottom to Top)

┌─────────────────────────────────────────────────────┐
│  AGENTIC SURFACES                                   │
│  CoWork (knowledge workers) ◄──► CoCo (developers)  │
└─────────────────────────┬───────────────────────────┘
                          │
┌─────────────────────────▼───────────────────────────┐
│  CONTEXT LAYER                                      │
│  Horizon Context + Cortex Sense + Semantic Studio   │
│  OSI standard (54 vendors) + Natoma MCP gateway     │
└─────────────────────────┬───────────────────────────┘
                          │
┌─────────────────────────▼───────────────────────────┐
│  SECURITY LAYER                                     │
│  Agent Identity + AI Security Posture Management    │
│  Cortex Guard + Data Exfiltration Policies          │
└─────────────────────────┬───────────────────────────┘
                          │
┌─────────────────────────▼───────────────────────────┐
│  AI INFRASTRUCTURE                                  │
│  Cortex AI (all frontier models + Cortex Training)  │
│  Adaptive Compute + Managed GPU clusters            │
└─────────────────────────┬───────────────────────────┘
                          │
┌─────────────────────────▼───────────────────────────┐
│  DATA FOUNDATION                                    │
│  Apache Iceberg V3 (GA) + Snowflake Storage         │
│  Datastream (Kafka) + Openflow (NiFi) + Postgres    │
│  SAP/Salesforce/Workday zero-copy integrations      │
└─────────────────────────────────────────────────────┘

Key Strategic Takeaways

1. Data is constant; models change — Snowflake's durable value is connecting the best AI model today to the enterprise data that is always there
2. Context is the new competitive moat — Cortex Sense (86% accuracy), Horizon Context, and OSI are bets that semantic context is more defensible than model quality
3. Security is the #1 barrier, now addressed — Agent Identity (GA) is the direct response to McKinsey's finding that 2/3 of enterprises cite security as the top blocker
4. Interoperability is a genuine commitment — Apache Polaris steering committee membership, real Fabric testing, OSI spec, 54 vendors — these are verifiable
5. Both audiences — CoWork for knowledge workers, CoCo for builders — ensuring all 13,900+ customers can access AI value regardless of technical sophistication

CEO's Framing

"The very nature of work is changing. Day-to-day work is about partnering with and guiding intelligent agents, and soon they will be operating across your business continuously, autonomously. This truly is going to be the era of the agentic enterprise."
— Sridhar Ramaswamy, CEO, Snowflake

---

Appendix A: Glossary

Term	Definition
Agentic Enterprise	Organization where AI agents work alongside humans continuously and autonomously
CoCo	Snowflake CoCo — AI coding agent for developers (formerly Cortex Code)
CoWork	Snowflake CoWork — Personal AI work agent for knowledge workers (formerly Snowflake Intelligence)
Cortex AISQL	SQL functions for running AI inference directly in SQL queries
Cortex Guard	ML-driven detection of prompt injection, jailbreaks, and zero-day AI vulnerabilities
Cortex Sense	Shared runtime context layer providing business definitions to all agents; 86% accuracy
Cortex Training	Snowflake-managed custom model training on GPUs; 2× efficiency vs. alternatives
Datastream	Snowflake's managed Kafka-compatible streaming service
Dynamic Tables	Declarative, auto-refreshing pipeline tables in Snowflake
Horizon Catalog	Snowflake's data governance and discovery layer, embedding Apache Polaris
Horizon Context	Extension of Horizon Catalog with Collect/Enrich/Activate for business metadata and semantic context
Iceberg V3	Latest version of Apache Iceberg with deletion vectors, row lineage, variant type, default values
MCP	Model Context Protocol — open standard for connecting AI tools to data
Natoma	Acquired startup providing enterprise MCP gateway with identity verification and audit controls
Openflow	Snowflake's managed data integration service built on Apache NiFi
OSI	Open Semantic Interchange — open standard for sharing semantic models; 54 participating vendors
SAP BDC	SAP Business Data Cloud
Semantic View Autopilot	AI-powered automatic creation and refinement of semantic views
Semantic Studio	UI for defining shared business logic without SQL expertise
Select Star	Acquired metadata intelligence company whose technology powers Horizon Context
ADE-Bench	Analytics and Data Engineering Benchmark by dbt Labs for evaluating AI coding agents
PEFT	Parameter-Efficient Fine-Tuning — technique for training large models efficiently
BYOC	Bring Your Own Cloud — architecture where data plane runs in customer's VPC

---

Report compiled from Snowflake official press releases, blog posts, product documentation, and media coverage of Snowflake Summit 2026 (June 1–4, San Francisco). Updated June 8, 2026 to incorporate all official press releases.

All feature availability statuses are sourced from official Snowflake press releases. GA = Generally Available, PR = Public Preview, PrPr = Private Preview.

Sources:

• Snowflake Press Release Hub: https://www.snowflake.com/en/news/press-releases/
• Atlan Summit Summary: https://atlan.com/know/snowflake/summit-2026-announcements/
• SiliconAngle: https://siliconangle.com/2026/06/02/ai-agents-open-data-governance-take-center-stage-snowflake-summit/
• Constellation Research: https://www.constellationr.com/insights/news/snowflake-summit-2026-context-custom-model-training-iceberg-v3
• Diginomica: https://diginomica.com/snowflake-summit-2026-how-snowflake-making-strategic-shift-towards-agentic-ai

---

SNACK-AI: The 2026 Pattern

As we look forward in 2026, SNACK-AI is the architectural pattern for powering AI applications at any scale - from mobile phones and robots to massive enterprise clusters.

SNACK-AI integrates:

• Snowflake Platform
• NiFi (Apache)
• Apache Iceberg
• Cortex AI (Agents/Search/MCP)
• Kafka (Apache Kafka)

Deep Dive into SNACK-AI:

• Read: SNACK-AI: The 2026 Pattern (Medium)
• Code: Official SNACK-AI GitHub Repository
• Weekly Context: SNACK-AI Overview

---

Thanks

https://sessionize.com/tspann

https://github.com/timothyspann