Python workloads in Snowflake can now use the package source your team already trusts. Customer-hosted Python Artifact Repositories, now in public preview, let you resolve approved packages for UDFs, stored procedures and Notebooks, with support for public HTTPS or PrivateLink connectivity. Check it out:

Deploying enterprise AI agents used to take months. Now it takes days. ⚡ @AtlanHQ's Prukalpa Sankar join Data Cloud Now to show how Atlan + Snowflake CoCo feed real-time business context without months of data movement. Watch here 👉🏻

❄️ The August Snow Report is LIVE ❄️ Two GA launches for Snowflake CoCo, warehouse tuning that handles itself, and #SnowflakeWorldTour# in 23 cities. 🚀What's new Snowflake CoCo Desktop is GA A native IDE inside your governed Snowflake environment, so it knows your data models and access policies from day one. Up to 2X faster on complex tasks with 51% fewer tokens than third-party assistants. Cloud Agents are GA Run agentic workflows from your browser in Snowsight. Start with a prompt, close your session, and the agent keeps working on Snowflake's managed infrastructure. Snowflake Adaptive Compute AI workloads are hard to predict. Adaptive Compute tunes performance and scale for you, so your engineers stop sizing warehouses by hand. Three CoWork features hit GA Mobile on iOS, Artifacts and Sharing, and Deep Research. 🗓️ What's ahead • Aug 18: Adaptive Compute in Action, faster workloads and simpler operations • Aug to Oct: Snowflake World Tour, 3 regions, 23 cities • On demand: Best of Summit 2026, 40+ top-rated sessions ➡️ Watch the full report and get everything in one place.

This just in: @SAP + Snowflake zero-copy integration is live on @googlecloud. Extend trusted SAP business context across your enterprise data platform, unified and governed, without moving a byte.

AI coding agents are stateless. Every session, same wrong turns. ArcticMem gives Snowflake CoCo persistent memory by extracting typed facts from successful sessions – the tribal knowledge nobody writes in a README. ADD/UPDATE/DELETE lifecycle keeps it clean. 47% → 73% overall pass rate. 18% fewer tool calls. Agents stopped chasing bugs that weren't bugs. 👇

AI guardrails are paving the way for safer, more creative human-AI collaboration. 🛣️ EDO's Kevin Krim shares how restricting AI data sets eliminates hallucinations and empowers human judgment. Watch 👉🏻

Ready to move your AI initiatives from experimentation to real-world impact? #SnowflakeWorldTour# is coming to 23 cities worldwide with practical strategies for building, deploying, and governing AI directly on your data. Here's what's in store: 🔹 Hands-on sessions on Snowflake Cortex 🔹 Generative AI for the enterprise 🔹 Data governance in the age of AI 🔹 Customer stories showing what's possible 🔹 Connections with local data leaders and experts Discover new ways to turn your data into intelligent applications! Register today for a city near you! 📍

Is your data ready for AI agents? Run /ai-readiness-score skill in CoCo CLI or Snowsight in Cloud Agents mode to get 0-100 score across consumption-ready tables, semantic view coverage and semantic view quality, plus a prioritized action plan. 👇

Unifying @SAP & Snowflake data is a game-changer for enterprise AI. 🤖 @Deloitte’s Christopher Dinkel & Gil Gomez break down how combining transactional data with governed management removes friction, feeding live data into AI models with total speed. Watch here 👉🏻

Data skew is not always about row counts. Snowpark lets engineers run Python, Java or Scala UDFs natively on Snowflake's query engine. At scale, two workers can process the same number of rows while one does far more compute. DySkew detects that at runtime and redistributes dynamically. No tuning required on your end. Result: ~30% lower P99 UDxF execution time in production, P90 down 27% on large warehouses. How we built it:

Migrating Apache Spark™ pipelines is more than rewriting syntax. With spark-migration in Snowflake CoCo, scan a codebase, flag unsupported patterns, apply fixes, and move to Snowpark Connect for Apache Spark™. Customers have seen up to 5.1x faster performance and 42% lower costs on Snowflake. Catch the demo below for more or check out this step-by-step guide:

Still treating warehouse migration as a multi-year project? Data Superhero Ruchi Soni covers Snowflake AIM: 94.7% automated conversion, 88% faster timelines, assessment-to-dbt in one workflow – bundled in CoCo CLI. 🔍 Read the full breakdown:

Scaling a company from 100 to 500 employees without your data flow collapsing isn't easy 🚀 @silverfort connected their entire data lifecycle using Snowflake and @boomi to build a clean, AI-ready data foundation. The result? Unified GTM data and improved sales and marketing efficiency as the business expanded. Check out their journey:

AI scales best when you empower your team to do more with the right tools 🛠️ Sumit Mehra from @TredenceInc joins Data Cloud Now to discuss AI Forge, an initiative equipping 5,000 employees with AI to boost productivity without shrinking the workforce. Watch here 👉🏻

Data engineering agents need harder evals than “write this SQL query.” data-eng-bench gives agents a live dbt project, an enterprise-scale warehouse, build/fix tasks and hidden verifier tests that check whether the resulting pipeline actually works. Now open source:

Need “about 497K unique users” faster than an exact COUNT(DISTINCT)? Postgres gives you a few options: TABLESAMPLE, HyperLogLog and mergeable sketches. The trick is knowing when each approximation works and where it breaks. Dive into this walkthrough:

AI agents need more than runtime guardrails. They need posture management. This demo shows AI Security Posture Management in Snowflake Trust Center: scan for AI-specific risks, assess config posture, and remediate misconfigurations before production issues show up. This is one of many updates in our latest AI Security capabilities 👇

More Snowflake CoCo users should not mean less cost control. Per-user quotas help teams scale AI-assisted development with clear usage guardrails and more predictable consumption. Watch the demo to see how it works or check out this step-by-step guide:

Unusable data architecture ruins AI strategies before they even launch. 🧱 Nectar360 overhauled their data foundations to ensure complete interoperability, unlocking seamless data clean room partnerships with Snowflake. Watch the full episode 👉

Snowflake CoCo is now listed on the NVIDIA Agent Skills GitHub repo: That brings NVIDIA skills into CoCo workflows, including cuDF, cuOpt, Nemotron and more. Reusable skills. Less custom scaffolding. More ways to build across Snowflake and NVIDIA. That brings NVIDIA agent skills into CoCo workflows, including skills tied to tools, libraries and models like cuDF, cuOpt and Nemotron. The useful part: less custom scaffolding when you want Snowflake-aware workflows to connect with NVIDIA’s accelerated computing ecosystem. Reusable skills. Familiar CoCo workflow. More ways to build across Snowflake and NVIDIA.