🎉 What's new

Your team already knows things about your data that can't be inferred from a column name: that this table refreshes at 06:00 UTC, excludes test orders, or that a null here means "not yet reconciled," not zero.

You may already have that context in Atlan as a Knowledge File.

But until now, the Context Agents that drive enrichment across so much of your data estate couldn't read it. They enriched purely from schema and lineage, with no way to point them at the document that already had the answer.

That changes now, with two new offerings.

First, link a Knowledge File to an asset (GA), and the Description and README context agents read it the next time they enrich that asset, grounding what they write in what your team already documented instead of inferring it. SOPs, best practices, data contracts, data dictionaries, runbooks, migration notes: all of this unstructured and semi-structured information can now shape how Atlan describes your data.

✨ Let's dig deeper

• Link from two places on the asset profile. Use the Knowledge files section in the details panel, or the + in the resources row under the Readme. Either opens the same picker: search, select one or more files, and click Link files.
• Scoped by design. An agent reads only the files linked to the asset it is enriching, never all of them. Linking is how you decide which document applies where.
• Written, not copied. Agents take the facts a document establishes and express them for the asset. You get the same facts in different words, not a trimmed passage.
• Nothing gets overwritten. Enrichment fills gaps. Descriptions and READMEs you already have stay exactly as they are.
• The document has to be a Knowledge File in Atlan. If yours isn't uploaded yet, create a Knowledge Folder and upload it first.

Second, and launching in Private Preview: Agent Templates

Everything above works in one direction: an agent reads a document you linked and writes onto an asset you already have.

Agent Templates run the other way. Instead of enriching what's in Atlan from schema and lineage, an agent reaches the systems where your documentation already lives, reads what's written there, and creates governed context that wasn't in Atlan at all - citing its source every time.

Each template is built for one outcome. You pick the outcome you want, connect the source, say what you're after, preview, then run. The first one is live:

Business Graph Term Generator turns your docs into business graph terms and categories, enriches terms that have no definition, and links them to the assets they describe. Every term carries a citation back to the document it came from.

• You supply four things, and none of them is the prompt. The source to read, the scope inside it, run instructions saying which terms you want, and optionally your own skills for house rules a template can't know. Atlan packages the rest: the skills that make a run reliable, the quality gates it has to pass, and the write path into Atlan.
• Sample before you publish. A sample run reads a small part of your sources and proposes a handful of terms in a minute or two. It publishes nothing. Read it, adjust your instructions, sample again, then run for real.
• Your Knowledge Files are a source here too. The documents you're linking to assets today can also be what a template reads from, with nothing to connect and no credentials to manage.
• Read-only by construction. For GitHub and Confluence, an admin connects the source once in Settings with a read-only credential you provide. That credential is the access boundary: the agent reads exactly what the token can read, can't write back to the source, and never asks an end user to sign in.
• More outcomes are coming. Selecting the bell on a template card records your interest, and that signal decides which one gets built next.

Agent Templates are available in the New UI only, enabled per tenant, and open to workspace admins and governance admins. Runs are free for the duration of the preview. If you don't see a Create agent button in Agents Studio, your tenant isn't switched on yet - talk to your Atlan contact!

New to this? Start with Understand agent templates.

🎉 What's new

Some work is better delivered as a standalone artifact than a chat reply.

You can now ask Conversational AI to create a standalone artifact from the context available in your workspace. Create formatted reports, dashboards, summaries, documents, diagrams, and more.

✨ Let's dig deeper

Create an artifact from a natural-language request

Describe what you need in your own words. When the answer is more useful as a deliverable, Conversational AI creates an artifact and shows it next to your conversation.

Use artifacts to bring together the information you need for activities such as:

• Reviewing governance or AI readiness
• Comparing assets and their business context
• Understanding lineage and downstream impact
• Summarizing findings for a team or stakeholder group

Read it beside your conversation

Select View on the artifact card to open the artifact in a side panel. You can read it while continuing to work in the conversation.

Refine the artifact with AI

Artifacts are designed to be iterated on. Select Edit with AI and describe what you want changed, such as adding an asset owner column, regrouping results, or adjusting the structure. Conversational AI creates a new version and refreshes the preview in place.

Download and take it anywhere

Download the artifact in its original format and use it wherever the work needs to go next - attach it to a ticket, share it with a teammate, or keep a snapshot before making further edits.

📄 Supported formats

Conversational AI can create:

• HTML for formatted dashboards, reports and summaries, including tables and charts
• Markdown for structured documents such as runbooks and documentation drafts
• Images for diagrams and other visual output
• Text for results that don’t need additional formatting
• Skills for reusable, governed capabilities. Describe the job, refine the draft, and save it as an Atlan asset. Learn more about skills and artifacts.

🔒 Built with privacy in mind

Artifacts are scoped to the conversation where they were created. HTML artifacts render in a sandbox and can’t reach external services or load external resources, so generated files can’t send your data outside Atlan.

👏 Give it a shot

Set up Conversational AI for your tenant, open a conversation, and ask it to create a report, summary, document, or diagram. For more on creating, editing, and downloading artifacts, see the Artifacts in Conversational AI guide. Try it and let us know what you think!

🎉 What's new

Atlan Lakehouse just got faster, more data and more reliable. Anything you need out of Atlan, from AI usage data to asset metadata to full exports, is available from the same open, governed layer, and is performant enough to build any application on. 

📊 Track how your teams use Atlan's AI interfaces

Atlan offers several conversational AI interfaces, so your teams can work with their enterprise context layer from any tool they want, whether it's their own agentic tools via Atlan MCP, Slack and Teams, or Atlan's built-in Conversational AI chat. You can now easily track how your teams use those interfaces directly in your own Atlan Lakehouse with two new tables:

The USAGE_ANALYTICS namespace in your Lakehouse now has two AI usage tables, refreshed daily: AI_TOOL_CALLS (one row per tool call across all of Atlan's AI interfaces) and AI_TURNS (one row per user question in Conversational AI, Slack, and Teams). Both sit alongside the metadata and product usage data already in your Lakehouse.

• Identify active AI users: Count questions and tool calls, and see your most active AI users by joining the new AI usage tables with USAGE_ANALYTICS.USERS.
• See which tools and clients lead: Use the client_name column to see which clients your teams use (e.g., Claude, Cursor, Codex, VS Code), and the tool_name column to see which Atlan capabilities they call most.
• Measure conversation depth: Group AI_TURNS by conversation_id ordered by timestamp to separate new questions from follow-ups, and understand how your teams' conversations evolve.
  
• Find friction fast: Use the is_error and duration_ms fields to see where AI calls fail or make people wait.

📊 Meanwhile, your asset exports just got dramatically faster

All asset export workflows (advanced, basic, admin) now run on a shared backend built on Atlan Lakehouse. It’s same UI, same configuration, same output but built on a faster, more scalable foundation. Exports that used to take more than a day now complete in less than half an hour, moving over a million assets. The switchover is behind the scenes – no changes needed from you.

👏 Get all the benefits of Atlan Lakehouse today

Atlan Lakehouse is generally available for all Atlan tenants.

• Use any engine: Pick up configuration details from Admin/Settings → Lakehouse (admin users only), and point any Iceberg REST-compatible engine at it, e.g., Snowflake, Trino, DuckDB.
• Use any AI tool: Install the /atlan-lakehouse skill to your favorite AI tool (e.g., Claude Code) and use natural language to ask any questions about your Atlan asset metadata and usage data – no query engine required.

And what might you want to access?

• Usage analytics: The USAGE_ANALYTICS namespace exposes page views, user actions, and adoption data you can analyze directly, with MCP and conversational AI logs now available.
• Relationship analytics: Metadata for all of your Atlan assets are in the Lakehouse, including user-defined relationships with full edge semantics, meaning your business-graph relationships are queryable for both your AI agents and analytical use cases, not just linked-asset lookups.

💥  More on the way!

Lakehouse already works with engines such as Snowflake, Databricks, BigQuery, Trino, and DuckDB. We're also building Atlan-managed Lakehouse Compute for when you want to query your data without running the compute layer yourself. More details to come!

🎉 What's new

As part of our ongoing improvements to the newly launched Atlan interface, you can now search your catalog or ask a question with a single command. 

Press ⌘K (or Ctrl+K) to open the Command center, one bar to search your catalog or ask Atlan AI. No more hunting through menus or remembering exact asset names. Just start typing.

✨ Let's dig deeper

✦ Search or ask in the same place
Type a question and Ask Atlan answers it conversationally; type a name and you get live asset results as you go. The Command Center figures out intent, so you don't have to switch tools to switch modes. Hit ⏎ to ask, ⇧⏎ to search.

🎛️ Use filters to narrow results fast
Start your journey by navigating the connection, database, schema, type and more to jump straight to what matters.

@ to tag, / to pick a skill
Type @ to pull a specific asset into your question, or / to choose an AI skill - then send it straight to a chat with the context already attached. From a vague idea to the right answer in a couple of keystrokes.

⌨️ Keyboard-first and accessible
Everything is reachable without the mouse — arrow keys, Home/End to jump, Esc to step back - and the whole experience is screen-reader friendly, announcing each result as you move.

👏 Give it a shot

Make sure you're on the new interface, then press ⌘K and start typing - search your catalog or ask Atlan, all from one place. We're rolling this out gradually and will keep refining it, so give the Command Center a try and let us know what you think.

Atlan now catalogs SAP Datasphere assets and traces their lineage across your estate. Available in Private Preview, this is the next step toward complete SAP Business Data Cloud coverage, joining existing support for SAP S/4HANA and SAP BW. It's also a step toward a context layer where SAP data is usable by AI agents wherever it lives.

What you can do

• Catalog SAP Datasphere views and analytical models alongside the rest of your estate, SAP and non-SAP alike.
• Trace context across systems. Follow a data product from its S/4HANA CDS-view origin, through SAP Datasphere, into your lakehouse and cloud platforms. Lineage doesn't stop at the SAP boundary.
• Preserve business meaning in motion. Join logic, filter rules, and business definitions travel with the data instead of being left behind at every hop.
• Make it usable by AI agents. Agents can discover the right SAP data and understand how it connects to Snowflake, Databricks, Google, and Microsoft, instead of guessing against raw schemas.

Why it matters 

For years, using SAP data for analytics and AI meant extracting it, copying it, and stitching it back together somewhere else. The join logic, the definitions, the meaning that made the data usable stayed behind. That's exactly what breaks AI agents built on SAP data: they inherit the extraction, not the context. As SAP opens Business Data Cloud to the broader ecosystem, Atlan connects the dots every time data leaves SAP and comes back. Context stays intact the whole way, so agents can reason over it. Increasingly, that enriched context flows back into SAP too.

Availability Private Preview, enabled per-tenant. SAP Analytics Cloud support is coming next to complete SAP Business Data Cloud coverage.

Setup guide: SAP Datasphere connector

🎉 What's new

Customers all over the world use Atlan MCP to build and manage their context layer, discover trusted datasets, enrich metadata at scale, and build autonomous data workflows, all from the AI tools they already use.

Over the past two months, we've shipped multiple releases to Atlan MCP that enable your AI agents to get better answers with tighter access controls, fewer tokens, and easier connectivity from more places than ever.

Here's what's changed:

🚀 Faster, smarter, leaner responses

- Search is now significantly faster: Search queries now return in seconds, powered by a new retrieval engine built for speed.

- Complex queries get better answers: Complex, open-ended questions are now routed to Atlan AI, allowing it to reason across your catalog, lineage, glossary, and data quality metadata, instead of requiring your MCP client to reason over raw catalog data itself.

- Responses use fewer tokens: Responses from Atlan MCP now use fewer tokens thanks to a summary page for agents, compacted responses, and only returning the most relevant connected subgraphs for lineage graphs.

- Smarter input handling: New safeguards against potential errors auto-correcting casing and type mismatches, normalize parameter names, auto-suggest fixes for mistyped or invalid values and make sure you get the answers you need.

🔒 Access controls that work everywhere

- Workspace-level MCP toggle: You can now gradually roll out Atlan MCP across your workspace. In Admin Center → Labs, you'll find a new toggle that you can use to enable or disable MCP for your workspace, and enable for everyone or admins only.

- User-aware results by default: Every interaction with Atlan MCP is properly scoped to user permissions and respects what the connected user's persona is allowed to see in Atlan, across both OAuth and API key sign-ins. Changes propagate immediately. 

🌎 Available in more tools

- Single connection URL: Connect to any Atlan workspace through mcp.atlan.com – one URL handles tenant routing automatically, no per-tenant configuration needed.

- More ways to get started: We've worked with ecosystem partners to make it easier to get started with Atlan MCP across more locations than ever:

• Claude Code plugin: Get started easily with Atlan MCP in Claude Code by installing the Atlan MCP plugin.
• ChatGPT App: Work with Atlan MCP in ChatGPT by installing the Atlan MCP ChatGPT app.
• Databricks marketplace: Use Atlan MCP with Databricks Genie by installing Atlan MCP through the Databricks Marketplace.

Atlan MCP is also supported across many other tools, including Gemini CLI, Snowflake Cortex Code, Claude Desktop, Cursor, Glean, Windsurf, VS Code, n8n, and Microsoft Copilot Studio.

View setup guides for all supported clients

👏 Give it a shot

Already using Atlan MCP? These improvements are now live, no action needed. Your existing connections benefit immediately from faster search, leaner responses, and tighter access controls.

New to Atlan MCP? Follow our documentation to connect your favorite AI tool to Atlan MCP.

🎉 Conversational Search now remembers you!

Ask about lineage in a new session, and it already knows which table you flagged as the certified source last week. No re-explaining, no re-tagging.

That's because conversational search now builds a persistent picture of you over time: your domain, the assets you keep returning to, and how you prefer answers structured.

✨ Let's dig deeper

• Memory is captured passively from conversations you're already having, no configuration required.
  
• Your existing conversation history seeds this from day one, so it isn't starting from zero, that continuity carries across every new session
• Your context stays yours. It doesn't surface in a colleague's session, and you can ask it to forget anything, anytime.

This is the start of something bigger. Today, this works inside Atlan. In the future, as you connect to Atlan through interfaces like MCP, that same context and history will travel with you, giving you and your agents the most useful information for any need.

Want to dig into the details? See more here

✨ Let's dig deeper

Your metadata estate is a knowledge graph: assets connected by meaningful relationships. With this release, those relationships, and the labels that describe them are first-class, queryable data in the Lakehouse, ready for analysis.

📊 Analytics on your ontology
Because relationship labels are now queryable, you can explore the shape and meaning of your estate in more creative and useful ways-

• Map your ontology - see every labeled relationship between terms ("Is a", "Derived by", "Allowed values") as queryable rows you can roll up and visualize.
• Measure coverage - find how many terms are connected versus orphaned, and where your ontology has gaps.
• Trace semantic lineage - follow "Derived by" and "Is a" chains to understand how concepts build on one another.
• Audit consistency - spot duplicate or inconsistent relationship labels and keep your ontology clean.
• Enrich AI & discovery - feed labeled relationships into the context that powers smarter search and conversational experiences.
• Combine and report - blend relationships with the rest of your Lakehouse metadata for richer, estate-wide reporting.

🌍 Where does this work?

• ✅ User-defined relationships used in the business graph
• This is a table called userdefrelationship inside the entity_metadata (also known as "raw" layer) namespace. You can join the table with other tables like glossaryterm to understand and map the relationships.

Why it matters: Your data estate is a knowledge graph, and its real value lives in the connections. By making relationship labels queryable, the Lakehouse lets you analyze not just what exists, but how everything relates  powering richer governance, discovery, and AI context.

👏 Give it a shot
Query the new relationship data, join it to your terms and assets, and watch your estate's ontology come to life.

Atlan for SAP: Ground Your Agents in the System of Record

● Generally Available

This release turns SAP's notoriously opaque data into a governed, machine-readable context layer for AI. SAP is the system of record for most large enterprises — and the hardest source to ground agents in, thanks to cryptic tables, coded field names, and configuration-driven logic. SAP ECC and SAP S/4HANA are now GA with two new capabilities that give agents (and the people who supervise them) two things they've never had over SAP: provenance — where a number truly comes from — and a map of the human layer, where SAP data is actually created and consumed.

CDS View Column-Level Lineage

Provenance

What's new: Column-level lineage is now generally available for SAP CDS (Core Data Services) views. Trace every column in a CDS view back through its transformations to the exact source table columns that feed it.

What you can do:

• Follow a single field end to end, from the semantic CDS layer down to the underlying SAP table columns.
• Run precise impact analysis — see exactly what breaks downstream if a source field changes.
• Debug and validate at the column, not just the object, level.

Why it matters for agents: This is the provenance backbone of the context layer. When an AI agent surfaces a metric, it can cite the precise origin and transformation path of every value — making the output auditable and trustworthy, and letting governance and quality signals propagate accurately across the SAP estate.

Fiori Apps as a New Asset Type

The Human Layer · First to market for Atlan

What's new: SAP Fiori apps are now a native asset type in Atlan, with asset-level lineage from Fiori apps → CDS views → upstream SAP tables.

The problem this solves: Fiori is the modern SAP UI — where business users actually work, reading, entering, and changing data every day, with near-zero visibility into the data beneath the screen. This release connects the app the user sees to the data it actually touches.

What you can do:

• See, for any Fiori app, the chain of CDS views and source tables it draws from.
• Give business users and stewards a clear map from the interface to the underlying data.
• Trace where sensitive or business-critical data is exposed and modified at the point of use.

Why it matters for agents: Fiori is the human layer of the context graph — where data is created and consumed. Mapping it lets an AI agent understand what an app does in data terms, and lets governance follow data all the way to the screen.

On the roadmap: column-level lineage for Fiori apps · popularity & usage signals for Fiori apps.

What's Next

Next: SAP Business Data Cloud (BDC). Zero-copy data sharing across Snowflake, SAP Databricks, Google BigQuery, and Microsoft Fabric. Context that follows your data across platforms without ever moving or duplicating it — so lineage, meaning, and governance stay intact wherever SAP data is consumed. Connectors for SAP Datasphere and SAP Analytics cloud, complementing our ERP and BW connectors.

More context ingestion from the SAP ecosystem, coming soon:

• Field-level help text as glossaries — SAP's own field documentation, automatically converted into governed business glossary terms and linked to the exact columns they define. The semantic layer that gives agents authoritative meaning for every SAP field.
• Master data as data products — your unique SAP configuration, packaged into governed data products: material types, customer and vendor account groups, and Business Partner groupings, roles, and categories. Context that reflects how your enterprise actually classifies its master data.
• SAP long text as knowledge files — SAP's free-text long texts (notes, descriptions, and documentation) harvested and published as knowledge files, ready to ground agents — for retrieval in agent studios and RAG workflows.

Beyond that — context from across the SAP application landscape (actively working with customers on outcomes and use cases for the below):

• SAP Signavio — business process context (how work actually flows).
• LeanIX — enterprise architecture and application portfolio context.
• SAP IBP — integrated business planning and supply chain context.
• SAP Concur — travel, expense, and spend context.
• SAP SuccessFactors — people and HR context.

The Bigger Picture: SAP + the Leading Non-SAP Context Layer

Pair the deepest context layer for SAP with the leading context layer for everything outside it — cloud warehouses, lakehouses, BI, transformation, and AI tooling — and the whole enterprise becomes legible to AI in ways neither side can deliver alone. An agent can trace a metric from a BI dashboard, through the cloud warehouse, across a zero-copy SAP share, into the CDS view and its source table — with provenance at every hop.

The estate is opening. The connective tissue is here. The result is a single, trustworthy map of how the business really runs — and the foundation for enterprise AI you can actually rely on.

🎉 What's new

Snowflake semantic views are now cataloged directly into Atlan as an open context layer home for the metric definitions that Cortex Analyst, Talk-to-Data and more agents depend on.

Atlan ingests the full hierarchy: the semantic view plus its logical tables, dimensions, facts, and metrics as discoverable, governable assets.

Because these views are the same objects Cortex Analyst runs on, cataloging them closes the gap between your governed metric definitions and the agents querying them.

And Atlan Context Studio lets you build and deploy new agents and semantic views from Atlan with the context they for agents that return accurate, governed answers

✨ Let's dig deeper

Here's what this looks like in practice and how it connects to Context Studio.

• Search for any semantic view and open its asset profile to see its logical tables, dimensions, facts, and metrics in one place.
• Feed these cataloged views directly into Context Studio, where they act as the primary execution surface. Context Studio either attaches to an existing semantic view or generates and updates its definition, with Atlan staying the source of truth.
• Apply the same governance you use elsewhere, like certification, ownership, README, and tags, to make a view trusted before it powers a production agent.
• Note: if your crawler role lacks the required grants, semantic views are skipped without failing the workflow, so an existing crawl won't break.

👏 Give it a shot

To start using semantic views:

1. Check the Snowflake permissions on your Atlan<>Snowflake connection
2. Run or schedule a metadata sync for your Snowflake connection.
3. Use global search and filter by asset type → Semantic view.
4. Open a semantic view to explore its Entities, Relationships, and Metrics tabs.
5. From there, point a Context Studio context product at the view to power an AI analyst.

📘 Full setup guide here. 

Finally, if you are at Snowflake Summit this week, come say hello and learn more about Snowflake+Atlan together!