The Context Layer, Live.
Everyone's talking about the context layer. Nobody's shown you how to build one. Until now.
Trusted by the world's leading AI-forward enterprises
Product Launches
The Context Engineering Workspace
AI bootstraps 90% of your context layer. Your team adds the rest — business logic, edge cases, knowledge only they have. Simulate. Refine. Ship to any execution engine.
Context Agents
9-12 months of enrichment work. Done in minutes. Nine AI agents solve the cold start problem the moment you ask.
Traces & Observation Loops
Agents get things wrong. Traces show exactly where. Corrections flow back into context automatically. Smarter every week — without touching the model.
Agenda
Enterprise Data Graph
The unified context layer for your entire data estate — connected by lineage, enriched by AI agents.
Unify Context
Context leaks at every system boundary. Atlan connects 80+ sources into one provenance chain — warehouses, BI tools, pipelines, AI systems. No silos. One unified view.
Enrich with Context Agents
Nine agent teams enrich at scale — descriptions, metrics, ontology, quality scores. 9-12 months compressed to 30 days. Stewards shift from documentation to certification.
Connect with Lineage
Column-level lineage across every boundary. Tag a column as PII once — governance propagates downstream. Change a schema — see the full blast radius across every dependent dashboard, pipeline, and AI agent.
Context Engineering Workspace
Engineer, test, and ship the enterprise knowledge that makes AI agents reliable.
Bootstrap Context Repos
AI bootstraps 90% from SQL patterns, lineage, and usage signals. Domain experts add the last 10% — business logic, exceptions, tribal knowledge. Versioned, bounded, portable to any execution engine.
Simulate & Test Context
Turn your BI dashboards into an automated eval suite. When a test fails, the studio pinpoints what’s missing and suggests a fix. Know when your agent is ready before your users find out it isn’t.
Observe & Improve
Every production interaction is a signal. Corrections become context updates. Correct answers become regression tests. Agents get more reliable every week — without the model changing.
Context Lakehouse
The first knowledge architecture built for a world where AI is the primary producer and consumer of context.
Knowledge Graph
Domains, semantic models, lineage, policies, and quality scores — all connected. An agent doesn’t just find an asset. It understands what it means, where it came from, and whether it’s trusted.
Iceberg-Native
The open table format your data lakehouse already runs on. Standard SQL. Full time-travel for compliance. No lock-in — your context is your IP. CME Group: 18M assets. Mastercard: hundreds of millions.
Accessible by Agents
MCP for governed queries. A2A so agents write context back. SQL for analytics. REST and Graph APIs for custom integrations. Every interface an agent needs — no tickets.
An open letter from our founders:
For months, the industry has talked about context.
We’ve given it many names: context graphs, semantic layers, systems of context. But for most teams, context has remained abstract, debated in blogs yet rarely operationalized.
From our inception in 2018, Atlan was never built as a catalog.
We were built around a broader idea: that metadata is context — and context is what allows teams to work together at scale.
That’s why, from the beginning, we went far beyond discovery. We built deep connectors not just into data platforms, but into operational tools, BI tools, communication systems, and developer workflows — because context doesn’t live in one place, and it doesn’t get used in one interface.
We helped enterprises connect lineage, ownership, and policies across some of the world’s most complex data estates. And we built the foundations for governance that actually works with data products, domains, policies, approval workflows, and more.
When the first wave of generative AI arrived in 2021, we saw what was coming.
AI wouldn’t just consume data. It would need to create, reason over, and act on context — continuously, and at scale. So we began rebuilding Atlan for that future: investing in platform primitives like an Iceberg-native Metadata Lakehouse, our App Framework, and the MCP server.
Atlan is now the unified context layer for humans and AI agents.
At Activate 2026, we’re showing — live — how the context layer actually gets built:
• The Enterprise Data Graph, and how Context Agents solve the cold start problem — compressing 9-12 months of enrichment to days
• The Context Engineering Workspace, to engineer, test, and ship the enterprise knowledge that makes AI agents reliable
• Traces & Observation Loops, so production AI behavior continuously improves context — without the model changing
• The Context Lakehouse — open, Iceberg-native, and accessible to every agent via MCP, A2A, SQL, and APIs
It’s time to take the context layer from theory to reality. See you at Activate!
— Prukalpa & Varun
Co-founders, Atlan
Join us, and the world’s leading businesses, to reimagine the future.
