Skip to main content

Search with conversational AI

Requires Atlan AI

Conversational search is only available if Atlan AI is enabled in your workspace. If you don't see it, ask your admin to set up conversational AI.

Conversational search lets you find data assets by asking questions in plain language. Instead of selecting filters or knowing the exact table name, you describe what you're looking for and get direct answers drawn from Atlan's metadata—including descriptions, ownership, certification status, data quality signals, and lineage.

Conversational search is available in several places:

  • Atlan homepage or chat tab: Open the chat panel from the Atlan homepage or the left navigation to search across your entire catalog.
  • Copilot on asset pages: Open Copilot from any asset page to ask questions anchored to that specific asset.
  • Your data tools: With the Atlan browser extension, conversational search is available directly inside Snowflake, Databricks, Power BI, and other tools.
  • Slack: Ask data questions in Slack channels once the Atlan app is added. See Integrate Slack.
  • Microsoft Teams: Access conversational search in Teams conversations. See Integrate Microsoft Teams.

Tips for better queries​

  • Name the asset type: Mention whether you're looking for a table, column, dashboard, or dbt model. "Find certified tables related to revenue" returns more focused results than "find revenue data."
  • Name the system or connector: "in Snowflake," "from Databricks," or "in the analytics schema" narrows the result set when you know where the data lives.
  • Use business terms: Conversational search understands glossary context. "revenue," "churn," and "MRR" work as search terms even if the technical table name is different.
  • Ask follow-up questions: If the first answer isn't quite right, refine it: "only show certified ones" or "filter to tables owned by the finance team." Context carries across the conversation.
  • Ask about quality in the same question: You don't need a separate search to check freshness or data quality. "Find the orders table and tell me if it has any staleness issues" combines both.

Example discovery queries​

Understand what data you have​

Give me a high-level map of our data estate: how many assets per connector, how many are governed (certified, owned, documented), and where are the biggest gaps?
Find all tables related to payments, transactions, or accounting. Break them down by system (Snowflake vs Databricks vs Postgres) so we can map our financial data landscape.
Find all tables related to revenue, billing, or invoicing across all our systems. Show which ones are verified and who owns them.
Compare these two assets: SALES_DAILY_AGG and SALES_MONTHLY_AGG and highlight differences in purpose, freshness, and downstream usage. What does the `churn_flag` column in SALES_DAILY_AGG mean?

Find data owners and experts​

I have questions about the MONTHLY_RECURRING_REVENUE table in Snowflake. Who owns it, and who else frequently uses it that I could reach out to?
A new VP of Engineering has joined and wants to meet our key data stewards. Who are the top 10 people who own the most assets? Show their names, asset counts, and which connectors they cover.
A team member is leaving next Friday. Find every asset they own—tables, views, dashboards, dbt models, glossary terms—so we can reassign ownership before they lose access. Their username is jane.smith.

Check data quality and freshness​

Find tables whose source was last updated over 6 months ago but are still certified, so we can flag them for review.
Are there any data quality issues flagged on @dim_products? Which tables in the analytics schema haven't been updated in over 7 days?

What conversational search can find​

Conversational search draws on all metadata stored in Atlan:

  • Tables, views, columns, schemas, and databases across all connected sources
  • Dashboards, reports, dbt models, Looker explores, and other BI and transformation assets
  • Business glossary terms and their definitions
  • Data owners, stewards, and subject matter experts
  • Data quality signals and freshness timestamps
  • Certification status, tags, and custom metadata

When to use structured search instead​

Use structured search when you need:

  • Precise filter-based navigation: filter by source, certification, owner, tags, or asset type in combination
  • Saved searches: bookmark a filtered view and return to it later
  • Shareable search results: share a filtered search URL with teammates
  • Browsing without a specific question: explore all assets of a given type or in a given schema

Next steps​