Your data.
Any shape.
In minutes.

The fastest semantic layer to deploy. Query governed metrics, build dashboards, and ship data apps from the AI tools you already use. All on reliable data.
Built in minutes from your terminal.

Read the docs →Free tier available. No credit card required.
claude code
❯ set up full semantic layer for me
✓ Explored 4 table schemas
✓ Read cube & view docs
✓ Created 5 cubes + 3 views
✓ Added descriptions to all fields
✓ bon validate — passed, no warnings
5 Cubes sales, online_sales, products, stores, customers
3 Views sales_overview, online_sales_overview, customer_360
✳ 2m 23s
❯ bon deploy
✓ Deployed. MCP server ready.
Problem

Imprisoned by tools that sold self-service.

Same questions every week. Dashboards nobody opens. Reports nobody trusts. And every AI gives a different answer to the same question.

Q3 RevenueAsk a Question
Q3 revenue by region?
AMER $2.4M
EMEA $1.8M
APAC $0.9M
WAU Trend
TotalBuild an Internal Tool
847
dashboards
RegionAMER ▾
metricvalue
revenue$2.4M
orders1,842
aov$1,303
NPS Score
Pipeline Q2Vibe and Deploy a Dashboard
AttributionCreate a Presentation
Q3 Revenue by Region
ARR
$4.2M
Solution

One definition. Every AI. Any output.

Your data team defines the metrics once. Everyone else gets reliable answers from the AI tools they already use, in whatever form they need. No new interface. No onboarding. No waiting.

Q3 RevenueAsk a Question
Q3 revenue by region?
AMER $2.4M
EMEA $1.8M
APAC $0.9M
WAU Trend
TotalBuild an Internal Tool
847
dashboards
RegionAMER ▾
metricvalue
revenue$2.4M
orders1,842
aov$1,303
NPS Score
Pipeline Q2Vibe and Deploy a Dashboard
AttributionCreate a Presentation
Q3 Revenue by Region
ARR
$4.2M
Features
01 — MCP Server

Five tools. Any AI assistant. One source of truth.

Explore schemas, query data, render charts. One endpoint that works everywhere your team already does.

ChatCoworkCode
revenue by category as a bar chart
Explore schema >
Bonnard MCPvisualize
$150M$100M$50M0
ElectronicsClothingHomeSportsBooksToys
Electronics dominates at ~$149M, followed by Clothing at ~$84M. Together those two account for the majority of total revenue. After that there's a steep drop-off to Home ($62M) and Sports ($41M).

Want me to drill into Electronics by subcategory to see what's driving that revenue?

Reply...
Opus 4.6
Claude is AI and can make mistakes. Please double-check responses.
MCP docs →
02 — DashboardsComing soon

Vibe code a dashboard. Deploy it.

Business users describe what they want in plain language. The AI builds it on your semantic layer. One command to deploy, accessible to anyone with permission.

ChatCoworkCode
build me a revenue dashboard by product category
Build dashboard >
Bonnard MCPvisualize
Revenue by CategoryQ4 2025
$150M$75M0
ElecClothHomeSportsBooks
$2.6M
Total Revenue
$142
Avg Order
Electronics
Top Category

Here's your revenue dashboard with a category breakdown and key metrics. Want me to deploy it?

yes, deploy it
$ bon dashboard:deploy revenue-by-category
✓ deployed to app.bonnard.dev/d/revenue-by-category

Done. Anyone with access can view it at that URL, or pull it up through MCP with @bonnard show revenue-by-category.

Reply...
Opus 4.6
Claude is AI and can make mistakes. Please double-check responses.
03 — Governance

Governed by default.

Metrics are version-controlled and deployed from the terminal. Access, roles, and row-level security are managed by admins from a dashboard.

bon cli
❯ bon diff
~ sales.revenue: SUM(amount) → SUM(amount) WHERE status = 'paid'
+ sales.refund_rate (new measure)
❯ bon deploy -m "Exclude unpaid orders from revenue"
✓ Deployed v4. 2 changes.
bonnard.dev / access control
Roles
marketingactive
Filter: department = 'marketing'
Cubes: campaigns, leads
financeactive
Filter: department = 'finance'
Cubes: revenue, costs, invoices
Read the docs →
How it works

Warehouse in. Governed metrics out.

Your data stays where it is. Bonnard adds a governed semantic layer on top. Every AI your team uses and your own code get reliable, per-user access.

Sources
Warehouses
SnowflakeSnowflake
BigQueryBigQuery
PostgreSQLPostgres
DatabricksDatabricks
DuckDBDuckDB
+10 more ways
Bonnard
Semantic Layer
Defined Metrics
Governance
Versioning
Metric Lineage
Apps
MCP
CopilotGemini
Ask Questions
Build Dashboards
Create Presentations
SDK / API
Internal Tools
Custom Data Apps
Embed Data
Setup

Your agent already knows how

bon init detects your stack, scaffolds your project, and writes agent context. Claude Code, Cursor, Codex. Ready in minutes, not quarters.

terminal
~ % bon init
Initialised Bonnard project
Core files:
bon.yaml
bonnard/cubes/
bonnard/views/
Agent support:
.claude/rules/bonnard.md
.claude/skills/bonnard-get-started/
.cursor/rules/bonnard.mdc
AGENTS.md
Claude Code
Cursor
Codex

No warehouse yet? bon datasource add --demo and you're exploring a full retail dataset in seconds.

Auto-detected from your project
Warehouses
Snowflake
BigQuery
PostgreSQL
Databricks
DuckDB
Data Tools
dbt
Dagster
Prefect
Airflow
Looker
Cube
Evidence
SQLMesh
Soda
Great Expectations

Have it ready to go before
your coffee gets cold.

Read the docs →