Wren AI
Wren AI is an open-source GenBI (generative BI) engine that turns plain business questions into governed SQL, charts, and deployable dashboards on top of the databases you already own. Underneath sits an open context layer — MDL — that gives both humans and AI agents a shared, machine-readable definition of what your data actually means: canonical tables, joins, calculated fields, enum values, and business rules stored as versioned YAML in git. A Rust semantic engine built on Apache DataFusion plans and executes modeled SQL across 20+ sources including PostgreSQL, BigQuery, Snowflake, Databricks, ClickHouse, DuckDB, Trino, Redshift, and SQL Server. Agents connect over MCP (Claude, Cursor, custom CLIs) or via LangChain / Pydantic AI SDKs, and can ship browser-side dashboards through wren-core-wasm to Vercel or Cloudflare Pages. The open core (CLI + MCP server + engine, no UI) is free forever for individual developers. The commercial layer adds GenBI Apps, agentic mode, dashboards, RBA
Analytics
Added on
Nov 4, 2026
Open sourceMCP serverText-to-SQLGenBI dashboardsMDL modeling20+ connectorsBring your own LLMCLI-firstSelf-hosted
Product information
Everything worth knowing about Wren AI
The complete picture — from everyday features to the technical detail builders and IT folks dig for.
Core features
What it actually does
GenBI Apps & Dashboards
Any AI agent can turn a business question into governed SQL and charts, then ship it as a shareable browser-side dashboard deployable to Vercel or Cloudflare Pages. Includes classic chat BI, AI spreadsheets, report caching and unlimited dashboards on every Cloud plan.
Unlimited Users, Usage-Based Billing
Unlimited members and projects on every plan — you pay for AI usage (credits on Cloud, concurrent sessions self-hosted), not seats. Your team and the agents calling Wren over MCP draw from the same pool, so company-wide rollout doesn't multiply license cost.
MCP Server & Embedded API
A first-class MCP server plugs Wren into Claude, Cursor, ChatGPT or any custom agent, plus an Embedded AI API for building analytics into your own product. Slack and Microsoft Teams integrations included; API calls carry no per-call fees.
20+ Data Sources, Queried In Place
Connect once to Snowflake, BigQuery, PostgreSQL, Databricks and 15+ more. Data is never moved or copied into Wren — queries execute against your warehouse in place, and columns you don't model stay invisible to the agent.
Technical capabilities
MDL Context Layer (Modeling Definition Language)
A YAML-based semantic contract defining models, relationships, calculated fields, primary keys and enum meanings, stored in files you can read, review, version in git and fork. Planning resolves modeled tables to physical tables via table_reference, strips undeclared columns, and applies business rules from knowledge/rules/ before returning rows.
Rust Semantic Engine on Apache DataFusion
A Rust engine plans and executes modeled SQL across PostgreSQL, MySQL, BigQuery, Snowflake, DuckDB, ClickHouse, Trino, SQL Server, Databricks, Redshift, Oracle, Athena and Spark. The same engine compiles to WebAssembly via wren-core-wasm so MDL-aware SQL and dashboards run entirely in the browser.
wren CLI, Skills & Framework SDKs
CLI commands for querying, planning, dry-plan validation, context building, data profiling and memory management. Structured Skills (generate-mdl, onboarding, enrich-context, genbi) let coding agents operate Wren reproducibly. LangChain (wren-langchain), Pydantic AI (wren-pydantic), WASM SDKs, plus dbt and OSI integrations.
Must watch videos
2:01
Tutorial2:01
Embedded Analytics with Wren AI: AI Assistants for Ecommerce, Healthcare EHR and Insurance
Embedded analytics with Wren AI: AI assistants for ecommerce customer support, healthcare EHR and insurance call centers.
Your users ask data questions everywhere: in your app, in chat, in their AI agents. Wren AI answers them all, from one governed engine, inside the product they already use.
In this demo, three products embed Wren AI so each user gets answers from their own data: an online store, a doctor's patient record, and an insurance agent desktop.
What you'll see:
- Retail support: a shopper thinks an order was charged twice. Wren checks the payments, finds one charge is only pending, shows the sources, and hands the case to support with the whole conversation attached.
- Clinical consultation: the patient record passes session properties (the doctor and the open patient), so Wren sees only this patient, and flags what needs a second look before a new drug, each with its source.
- Insurance claims and quotes: Wren answers only within the agent's role, the policy and the claim, quotes roadside assistance mid-call, and compares claims as totals only, never other customers' records.
- Three ways in: Embedded Threads puts the full conversation in your app with one snippet. The Embedded AI API lets you build your own interface. Over MCP, Claude, ChatGPT, or your own agent can call Wren directly.
Three ways in. One governed engine.
Chapters
0:00 Intro
0:13 Retail: customer support assistant
0:37 Healthcare: clinical consultation co-pilot
1:01 Insurance: claims and quote agent
1:27 Three ways in: Embedded Threads, Embedded AI API, MCP
🌐 Try the live demos: https://getwren.ai/embedded
📅 Book a 30-min demo: https://calendly.com/d/d3hn-4h3-pw4/30-min-product-intro
The demos use sample data, for illustration only. In the healthcare demo, Wren AI retrieves and summarizes existing records; it does not diagnose, prescribe, or give medical advice. Quotes in the insurance demo are illustrative, not offers of coverage.
#WrenAI #EmbeddedAnalytics #AIAssistant #WhiteLabel #GenBI #mcpe2025
0:37
Tutorial0:37
White-Label AI Analyst for Your Trading Dashboard
See how Wren AI embeds a conversational analyst straight into a capital markets dashboard. No new dashboard to build.
In this 35-second demo, a user looks at a Costco stock chart, opens "Ask Wren AI," and asks: "Which stocks in my watchlist gained the most over the past year?" Wren AI answers using the dashboard's own data.
What you'll see:
- Embedded analytics: explain what moved a stock, right next to the chart
- Session context: the agent already knows your watchlist and the chart on screen
- Context layer: terms like "gained the most" are defined once in the semantic layer (MDL), not guessed
- Traceable answers: every number comes with the underlying data and the SQL behind it
- Instant charts: turn results into a visual, like the 1Y return by symbol bar chart
⭐ Book a demo: https://www.getwren.ai/en/request-demo
#WrenAI #EmbeddedAnalytics #CapitalMarkets #AIAnalytics #TextToSQL #FinTech #SemanticLayer #GenerativeBI
0:37
Tutorial0:37
Build AI agents that never forget
Tired of explaining who you are and how you like answers every time you start a new chat? With Memory in WrenAI, you say it once and Wren remembers it for every new chat in your project.
In this video you'll see:
- Why every new AI chat normally starts from zero
- How Wren Memory saves your role, preferences, corrections, and what your words mean
- Why your Memory is private to you, and how it differs from Knowledge, which your whole team shares
✨ Try Wren on your data: https://cloud.getwren.ai
🌟 Book a demo: https://calendly.com/d/d3hn-4h3-pw4/30-min-product-intro
Who uses it
Real use cases, no hypeWhat's great
- Free open-source engine — audit every line
- 20+ sources queried in place, no data movement
- Unlimited users on every plan — pay for usage, not seats
- MCP server drops straight into Claude and Cursor
Technical strengths
- Rust semantic engine on Apache DataFusion plans modeled SQL
- Dry-plan validation catches broken SQL before the warehouse
- wren-core-wasm runs MDL-aware queries in the browser
Where it falls short
- Open source is CLI/MCP only — zero UI
- Credit pricing burns fast on multi-step agent runs
- Classic GenBI app is sunset — docs still confuse people
- Self-hosted requires a sales contract, no self-serve
Technical limitations
- Manual MDL modeling is heavy upfront work
- Operational and behavioral context layers still unfinished
- legacy/v1 branch gets no security fixes
For technical folks
The deep-dive specs
Semantic engine
Rust engine powered by Apache DataFusion
Context format
MDL (Modeling Definition Language) — YAML files, git-versioned
Interfaces
wren CLI, MCP server, Embedded AI API, LangChain, Pydantic AI, WASM
Browser runtime
wren-core-wasm (DataFusion engine compiled to WebAssembly), deployable to Vercel / Cloudflare Pages
Supported data sources
PostgreSQL, MySQL, BigQuery, Snowflake, DuckDB, ClickHouse, Trino, SQL Server, Databricks, Redshift, Oracle, Athena, Apache Spark (20+)
LLMs offered on Cloud
gpt-5.6-luna (default), gpt-5.6-terra, gpt-5.6-sol
Deployment options
Cloud, self-hosted VPC, on-prem, air-gapped
License model
Open core — engine open source; platform & enterprise ops commercial
Compliance
SOC 2
Context layers
Structural & semantic & business (shipped); operational & behavioral (in development)
FAQ
Includes technical Q&AOpen source: the context engine for individual developers via CLI and MCP (no UI) — MDL modeling, core text-to-SQL pipeline, and 20+ connectors. Commercial: the advanced AI harnesses (GenBI Apps, agentic mode, AI-assisted context prep), the UI and dashboards, embedded APIs, RBAC, row/column-level security, and support, delivered as Cloud or self-hosted Enterprise Plus.
Traffic Insights
Monthly Visits
—
November 2026
Growth Rate
0%
vs last month
Dominance
0%
in Analytics
Top Country
—
Top Countries Breakdown
—
Growth Trend
Stable
Traffic has remained stable.
Honest pricing
No sneaky tiers, no “contact sales”
Here's who each plan is actually for — and where the hidden charges might hit.
View Full Plans & Pricing on Wren AI
Official pricing, checkout, and plan upgrades are handled directly on getwren.ai.
Reviews
What the internet actually thinks
Data refreshed weekly. No paid placements.
Trustpilot
Not reviewed yet
G2
Not reviewed yet
Capterra
Not reviewed yet
Product Hunt
Not reviewed yet
COMMUNITY COMMENTS
Leave a Comment
Help others make informed decisions. Your honest feedback shapes the community's understanding of this tool.
Tips for commenting
- •Be specific about features you used
- •Share real use cases and results
- •Mention both pros and cons
- •Keep it honest and constructive
to leave a comment
Your comment is published immediately.
Sort:
No comments yet
Be the first to share your experience with this tool.
Alternatives
Not quite the right fit?
Here's what else is out there.
