Try free

Embedded GenBI

Ship conversational analytics inside your product.

Embed Wren's governed data agent, white-label, through Threads, APIs, or MCP, on any of 20+ databases. One context layer governs every answer. Your brand, your policies, live in days.

Try Wren AI Cloud freeRequest a demo
  • Embedded Threads
  • AI API
  • MCP
  • Slack and Teams

Trusted by data teams worldwide, with 17,793 GitHub stars

Inside a real product

Your users ask. Your platform answers.

Wren AI runs as the analyst inside your product, answering from your live tables, in your brand, scoped to whoever is signed in. Pick a question to see the exchange.

Trading intelligenceEmbedded Threads
sample product
Trading Intelligence
SPX5,812.4+0.41%NDX20,488+0.63%VIX15.8−3.1%DXY101.2+0.05%US10Y3.94%−2bpWTI68.31+0.88%GOLD2,661+0.22%BTC64,120−0.74%SPX5,812.4+0.41%NDX20,488+0.63%VIX15.8−3.1%DXY101.2+0.05%US10Y3.94%−2bpWTI68.31+0.88%GOLD2,661+0.22%BTC64,120−0.74%

Overview

Overnight session · Sep 29 18:00 → Sep 30 09:30 ET
Overnight P&L
+$48,210
Decisions / executed
37 / 29
Win rate
68.9%
Max drawdown
−0.21%
  1. Context layer

    “Overnight” means one thing

    Session windows, P&L, win rate, and breadth are defined once in MDL, so every answer uses the platform’s own definitions instead of guessing at column names.

  2. Signed JWT

    Each desk sees only its own book

    The host signs a token per user. Wren AI turns it into row-level filters inside the engine, so a PM on desk-07 can never query another desk’s fills.

  3. White-label

    Your brand, our agent

    The panel carries the platform’s colors and name. Answers come with the trades and the SQL behind them, so users can check every number.

Sample product and data, for illustration only. Answers are read-only and describe what already happened; Wren AI does not place trades or give investment advice.

Choose your surface

Three ways in. One governed engine.

Every surface resolves through the same context layer: same metric definitions, same access policies, same audit trail.

index.htmliframe
<iframe
  src="https://cloud.getwren.ai/iframe/your-unique-id"
  width="100%"
  height="100%"
  style="border: none;">
</iframe>

Drop the full conversational experience into your app with one snippet. End users ask questions in plain language; Wren AI returns governed SQL, charts, and summaries under your brand.

Embedded Threads docs
  • Public mode, or JWT identity mode signed server-side (HS256)
  • Per-user data scoping via sessionProperties in the token
  • Brand the chat icon, company logo, and primary color

Embedded Threads

Conversational analytics in your app, in an afternoon.

Embed the full Wren AI thread experience with an iframe. Your users explore their own data in plain language without leaving your product, and every answer stays inside your governance boundary.

Read the docs
A white-label Wren AI chat thread embedded in a host application answers a question with a governed chart

Embedded AI API

Your UX on top. Wren's data agent underneath.

Two endpoint families cover the whole loop: natural language to SQL with execution, and chart generation as Vega-Lite specs. Conversation context carries across follow-up questions.

Explore the API reference
A curl request to the generate_sql endpoint flows through the Wren engine and returns SQL with a chart

MCP, Slack and Teams

Meet your users in the tools they already open.

Wren AI speaks MCP natively and ships Slack and Teams integrations. Agents and teammates query through the same definitions your product embeds; nobody touches raw tables.

Read the MCP docs
Claude, ChatGPT, Slack, and Teams all query one governed context layer with RLS, CLS, and audit logging

How it ships

Model once. Ship everywhere.

The context layer is the work; the surfaces are configuration.

  1. Step 1

    Model once

    Define metrics, relationships, and business logic in MDL. Connect 20+ sources, with dbt sync included.

  2. Step 2

    Pick your surfaces

    Embedded Threads for speed, the AI API for control, MCP for agents, Slack and Teams for reach.

  3. Step 3

    Ship white-label

    Apply your logo, color, and chat icon. Sign JWTs for per-user scoping. Go live in days, not quarters.

Multi-tenant by design

Every tenant sees exactly what they should. Provably.

Embedding analytics means putting your customers' trust on the line. Wren AI enforces the boundary in the engine, not in your application code.

request traceenforced in engine
  • sessionjwt.verify(HS256)valid
  • tenantsolstice, user 4821scoped
  • policyrow_level = tenant_idapplied
  • policycolumn_mask(email)applied
  • auditq_8f21, 1 row returnedlogged
Identity

Signed JWTs

Your backend signs every embedded session with a shared secret. The key never reaches the browser.

Scoping

Row-level and column-level controls

Access policies are enforced inside the engine at query time, per end user.

Audit

Full activity logs

Provable who asked what, and which query produced which number.

Isolation

Projects and per-app tokens

Separate tenants by project. Issue one API token per surface and revoke it independently.

Estimate build vs. buy ROI

Months to build. Days to embed.

Building an enterprise-grade text-to-SQL context layer in-house takes months of engineering and constant maintenance. See how much you save on development cycles and cloud margins by embedding Wren AI.

Request a demoCalculate your GenBI ROI
Wren AI