📰 All Data and AI Weekly #230-23 Feb 2026

#llm #mcp #agenticai #ai

📰 All Data and AI Weekly #230-23 Feb 2026

( AI, Data, NiFi, Iceberg, Polaris, Streamlit, Flink, Kafka, Python, Java, SQL, MCP, LLM, RAG, Cortex AI, AISQL, Search, Unstructured Data )

Blizzard!!! So much snow here.

🗓️ Upcoming Events: Catch Tim Live!

March 6: LLM Days NYC - NYC Tim Speaking
March 11: MCP Conference NYC – NYC Tim Speaking
March 12: Data for Breakfast – NYC Hands on Labs
May 20: Budapest Data and AI – Budapest, Hungary - Tim Virtual speaking

🌟 Featured Highlight: Tim Spann’s Corner

IoT meets Agentic Streams

This week, Tim Spann continues to push the boundaries of real-time data and agentic integration. Two major repositories caught our eye:

SNACKAI-CoCo-MCP-Server-Integration: A cutting-edge implementation focusing on Model Context Protocol (MCP) server integration. This is a must-watch for anyone looking to build standardized interfaces for AI agents to interact with IoT data streams.
Snowflake-Initializr: Simplifying the "Day 0" experience for Snowflake developers. This tool streamlines the setup of new projects, ensuring best practices are baked in from the first line of code.

❄️ The Snowflake Deep Dive

Snowflake is moving fast on two fronts: Governance/DevOps and Agentic Intelligence.

Infrastructure & Governance

Version Control for Semantic Views: Snowflake is bringing software engineering rigor to data modeling. Read more on the Engineering Blog.
Terraforming Snowflake: A comprehensive guide to managing your data warehouse as code. Check out the Quickstart.
Cost Management & Anomalies: With new documentation on Cost Anomalies and insights into spending patterns, keeping your cloud spend under control has never been more visible.
Trust Center Integration: Learn how to Classify UI and leverage the Trust Center to secure your data estate.

The AI Frontier

Cortex AI + Claude 3.5 Sonnet: The powerhouse model is now integrated into Snowflake Cortex AI, enabling high-reasoning tasks directly within your data platform. Read the announcement.
Cortex Agents & Web Search: Snowflake agents can now go beyond your internal data to perform Web Searches.
Observability: Monitor your agentic workflows with the new Cortex Agent Usage History.

🤖 Agentic Ecosystem & Research

The industry is moving toward "World Models" for agents. Here are the key resources released this week:

Snowflake Agent World Model: Snowflake Labs has open-sourced the code and datasets (1K) for training agents that understand complex environments.
MIT AI Agent Index: A new benchmark and directory for tracking the capabilities of the world's leading AI agents.
AgentSkills.io & Vercel Skills: Frameworks for defining and deploying the specific capabilities (skills) that agents use to interact with the world.

🛠️ Developer Tooling & Data Movement

Netflix Data Bridge: A fascinating look at how Netflix simplifies massive data movement across their infrastructure.
Liquibase Secure 5.1: Change control for Snowflake is getting a security boost.
PostgresGUI.com: A new, streamlined way to manage your Postgres instances.
WebMCP: Bringing the Model Context Protocol to the browser.

Notable Repositories

AirLLM: Run large models on consumer hardware with extreme memory efficiency.
OpenObserve: A cloud-native observability platform that’s gaining serious traction.
Nixtla: The go-to library for state-of-the-art time series forecasting.

📈 Careers & Growth

From BI to AI: Jimish Kadakia shares a roadmap for evolving from a BI Developer to a Solutions Architect in the age of Snowflake and AI.

Stay curious, keep streaming.

🍭 2026: SNACK-AI

As we look forward in 2026, we are introducing SNACK-AI. This architectural pattern is designed to power AI applications at any scale—from mobile phones and robots to massive enterprise clusters.

SNACK-AI integrates: