SC26Meet Wren AI at SC26 Chicago, booth #5413, November 17–19

The governed GenBI agent

Ask your data.
Get one trusted answer.

Your team asks in plain language. Claude, ChatGPT and your own agents ask over MCP. Everyone gets the same answer, built on business definitions you control.

18,303stars on GitHub
Governed Context Layer
GenBI + MCP

Trusted by data teams worldwide

See it work

Type a business question in plain English. Wren builds a dashboard on your data that you can drill into and filter.

Wren AI
HC
Campaign Strategy dashboard

ASK WREN AI

|

Generate
Reading your prompt

How it works

Your data, queried in place
  • Snowflake
  • BigQuery
  • Databricks
  • Redshift
  • PostgreSQL
  • SQL Server
  • Oracle
  • MySQL
  • ClickHouse
  • Athena
  • Trino
  • Starburst
Wren AI
Same answer, everywhere
  • 01

    Connect your data

    Snowflake, BigQuery, PostgreSQL and 20+ more. Your data stays where it is, with no ETL or migration.

  • 02

    Teach it your business

    The onboarding agent drafts your context layer: metrics, relationships and definitions. Your data team reviews and approves it.

  • 03

    Ask from anywhere

    Your team asks in Wren, Slack or Teams. AI agents ask over MCP. Everyone gets the same answer.

Why teams trust it

Every answer uses your context layer, respects your access rules and shows its work.

Shows its work

Every answer comes with its reasoning and the SQL it ran.

SANDBOX, ISOLATED RUNTIMERUN #4821, REPLAYABLEWhich regions drove Q3 margin?context.lookup"gross margin"margin = (revenue − cogs) / revenuesql.runvalidated12 rows, 0.8 s, matches ground truthchart.barregion, marginrendered to artifactskill.save"q3-margin-by-region"reusable on the next askARTIFACTNAEUAPACLATAMMEASKILLTRACEEVERY STEP REPLAYABLE
Respects your access rules

People and agents only see the rows and columns they're allowed to.

MCP, Claude, role: analyst-EMEAApp, CFO, role: finance-allAppAPISlack/TeamsMCPONE POLICYregionrevenuessnEMEA1.2M412-88-1937•••-••-1937APAC0.9M305-41-8820•••-••-8820EMEA0.7M118-62-4473•••-••-4473NA1.5M527-09-3316•••-••-3316EMEA0.4M260-77-5190•••-••-51903 of 5 rows, ssn masked5 of 5 rows, ssn maskedAUDIT LOGanalyst-EMEA3 rows, maskedvia MCP, 14:02cfo5 rows, maskedvia App, 14:02

Customers

See all industries
Data shifted from an "expert tool" to a "company-wide asset."
Shasta HoCEO, Nextlink

FAQ

Wren AI, in plain terms

Wren AI is an enterprise AI analytics platform built by Canner, Inc. Your team asks questions in plain language, and AI agents such as Claude and ChatGPT ask over MCP. Both get governed answers from your existing databases: validated SQL, charts, dashboards and GenBI Apps, grounded in business definitions you control and limited to what each person is allowed to see.

The context engine underneath Wren AI is open source on GitHub: a CLI and MCP server for an individual developer, with no UI. Wren AI for teams and companies is a commercial product, sold as Wren AI Cloud or self-hosted Enterprise Plus, with vendor support and an SLA. Both run the same engine, so your definitions carry over and there is no lock-in. See open source vs. commercial.

The web UI and GenBI agent for business users, dashboards and GenBI Apps, agentic mode with Skills and Memory, AI-assisted context preparation, Slack and Microsoft Teams, embedded analytics APIs, SSO with user and group management, row- and column-level security, audit logs, managed or self-hosted deployment, and vendor support.

Data and AI leaders at mid-market and global enterprises with many data sources, a backlog of ad-hoc questions, and strict governance needs, plus the data architects, analytics engineers and platform teams who run it. Software companies also embed Wren to ship AI analytics in their own products.

Every question is planned through a context layer that defines your metrics, joins and business rules, kept as MDL files in your own git repo, instead of letting an LLM guess from raw table names. Wren validates the SQL before it runs and applies row- and column-level security for the person or agent asking, at query compilation rather than with prompt filters. Agents never connect to your database directly, and every query is logged. See security and governance.

Choose Wren AI Cloud, your own VPC, on-premise, or fully air-gapped with local LLMs. In every option your tables stay in your databases and are queried in place. Self-hosted customers pick their own model endpoint, so query results can stay inside their network. Compare options on Cloud vs. self-hosted.

Every plan includes unlimited users. Wren AI Cloud starts at $179 per month billed annually, with a 7-day free trial. Self-hosted Enterprise Plus is licensed by concurrent sessions and quoted by our team. See pricing.

BI copilots such as Power BI Copilot or ThoughtSpot keep your logic inside one BI tool. Text-to-SQL tools let an LLM guess joins and metrics from your schema. Standalone semantic layers need a separate BI and AI front end. Wren gives you the whole stack, from the context layer to the agent, dashboards and an MCP server, and serves the same governed definitions to every person and AI agent. See Wren AI vs. alternatives.