Atlan

Your AI agent found the data. But should this user see it? In this demo from Atlan Activate, Steven Touw, Co-Founder & CTO at Immuta, walks through what governed agentic access looks like end to end. An analyst discovers a patients table in Atlan's context layer, requests access, gets approved, and queries the data in Snowflake — without a single manual ticket. Immuta pushes native policies directly into Snowflake the moment approval happens. Then it gets interesting. 👀 The same workflow through an AI agent. A user asks a question through Claude, Atlan surfaces the relevant tables over MCP, and Immuta vends a scoped role on the fly. Zero standing permissions. Masked columns stay masked until a break-glass exception is approved. Policies update in real time. Atlan knows what data exists. Immuta knows who's allowed to see it. Together, governed access happens on the fly. Check out the clip. Full demo in the comments.

Let's stroll before Snowflake. 👟 🌉                                                                                            Grab your sneakers and us on June 2 for Context in Motion — a casual morning walk through San Francisco with fellow data and AI leaders. Fresh air, great conversations, and meaningful connections before the showfloor opens.                                                                                                             Open to all. 6:30 - 8:30 AM PT. RSVP: https://lnkd.in/dcrKfHve

There's a difference between context for data and context for AI. Most enterprises haven't figured this out yet. Great conversation with our cofounder Prukalpa ⚡and Stratola LLC digging into what that actually means in production.

At Activate 2026, we stopped talking about what context could do — and showed what it already does. Here's what went live: 🔹 Context Agents collapsed a 9–12 month metadata enrichment journey into 30 days. 50 teams created over 1 million AI-generated descriptions in two weeks — a 40x output increase with a fraction of the headcount. 🔹 Context Engineering Studio took an AI agent from blank page to production deployment in under 10 minutes — built live in Cursor, pulling skills, semantic models, and SOPs directly from Atlan's context layer. 🔹 The Context Lakehouse logged 8 billion reads in 90 days. One context layer, readable by every execution engine — Cortex, Genie, Claude, Codex — via MCP, API, or SQL. The customers in the room said it best: "For years, we've been taught that documenting tribal knowledge has to be manual. Context Agents proved you can lead with technology." — Governance Lead at Engine "Any team, any tool, same answer is what we want to drive." — Jessie Buelteman, Elastic Watch the highlights ⬇️

On a track, what's underneath the hood is what wins. Production AI works the same way. Atlan, Hakkōda, an IBM Company, and Sigma co-hosted Snowflake's Southeast and Northeast sales leadership at the BMW Performance Center for a full day on what it actually takes to get AI workloads from pilot to production. Days like this build the kind of partnership that lasts. Big thanks to Matthew Moscoffian, Adrian Tarquinio, and Rob Silva for bringing your Snowflake teams and the energy that made the day. Hakkoda and Sigma showed up as the co-hosts every partnership wants. Shoutout to the Atlan pit crew on-site, Charlie Freeman, Razi Shafiq, Jon B., Tom Linton, and Brantley Berryhill who kept the wheels turning.

Happening today! ⏰ Come see how Databricks Genie, grounded in Atlan's context layer, powers AI agents that reason from trusted data and explain every answer. 👉 https://lnkd.in/gCNfRG-Z                                 

Meet Maya. A customer support executive. She succeeds because she has context, experience, and the right knowledge. Enterprise AI systems need the same thing: business knowledge, historical signals, decision frameworks, tools, and human judgment. For years, organizations invested in moving data to the cloud, building analytics layers, training smarter models. The models got exponentially better. Enterprise AI didn't get exponentially more useful. The real challenge was always context. Can your AI understand your business? Interpret institutional knowledge? Reason with trust, lineage, and business logic? Act with the right guardrails? The next generation of AI platforms will win by connecting metadata, understanding lineage, learning business semantics, and enabling trusted AI agents. The future of AI is contextual intelligence. ✨

Snowflake Summit is in two weeks! ⏳😯 We're bringing the full context layer conversation to San Francisco. Four sessions. One spicy debate. Workday on stage sharing what they've built in production. And Prukalpa ⚡ demystifying the enterprise context layer live. What we're tackling: → WTF is the enterprise context layer, actually? Prukalpa breaks it down and demos it. June 1, 4:00 PM PT. → Context layer vs. semantic layer vs. ontology vs. knowledge graph. Three industry leaders, one stage, no agreed-upon answer. Prukalpa Sankar, Jessica Talisman, and Cindi Howson hash it out. June 2, 1:00 PM PT. → Workday's VP of Enterprise Data & Analytics shares how they're putting the context layer into production. Not theory. Proof. June 2, 3:00 PM PT. → Workday's Director of Enterprise AI Solutions goes deeper on building trustworthy AI agents with context. June 3, 12:30 PM PT. Come see the Context Layer LIVE at Booth #1612 all week. See you in SF 👇

Builder Demos ✨🚢 See what Atlan shipped this week Every Thursday, our teams demo what they're building with AI. Here are three from this week's builder demos: ➡️ Mirijam Stewart celebrated her one-month anniversary at Atlan by demoing a Dungeons & Dragons-style training simulation for new customer support team members. Her skill pulls real customer data, generates realistic account scenarios with probability-based challenges, and lets CSMs role-play through them before they ever get on a real call. ➡️ Arathy Akkamadam built a system that uses a website claw to auto-audit our website every week, generates experiment hypotheses backed by real traffic data, creates the variants, and tracks results. The team went from running 1-2 website experiments a month to 10. ➡️ Dubem Izuorah showed the impact of an autonomous agent that monitors website health, finds JavaScript errors, checks browser consoles, and opens PRs to fix performance issues. Results that would normally require full sprints are now shipping through merged PRs. This is what AI-native looks like in practice. Not a strategy deck. A CSM, a marketer, and an engineer building and shipping every week.

We're heading back to San Francisco for Databricks Data + AI Summit, June 15–18.  🙌                                                                                                                               Two sessions on stage, including Mastercard presenting how they built context by design on Databricks & Atlan. Live demos at the booth. And a full week of events for the leaders building enterprise AI infrastructure.                                                                                     Your agents don't fail because of the model. They fail because they don't know what your company knows. Come see the fix.                                                           Book a meeting → https://lnkd.in/dsDpFtyb