Open source · Apache 2.0
The open framework for agents to build your own BI
anfra is an open framework for building your own AI-native BI platform: flexible for your team, governed through a semantic layer
curl -fsSL https://anfra.ai/install.sh | bashCustom BI tools
Design your own data apps BI experiences
Ask your coding agent for any tool your team needs, from a metric explorer to something no BI vendor ships, all built on the same governed metrics
- No filtersMeasures2Orders1,284Revenue$86.4kAdd a measureShowingOrdersCountry# ↓United States486Vietnam412Germany228Japan158Category# ↓Software538Hardware431Services315Channel# ↓Organic602Paid search389Referral293Plan# ↓Pro640Team402Free242Device# ↓Desktop701Mobile452Tablet131Customer# ↓Returning776New508
Build me a metric explorer where I can pick any measure, filter by a leaderboard row and compare to the previous period.
- Row ($K)Q4 2025Q1 2026Q2 2026Q3 2026TotalOperating activitiesCash receipts1,2401,1851,4021,4685,295Payroll−612−630−655−671−2,568Vendor payments−284−301−296−318−1,199Net operating cash3442544514791,528% change in Cash receiptsvs prev quarter+4.2%−4.4%+18.3%+4.7%–% change in Net operating cashvs prev quarter+5.8%−26.2%+77.6%+6.2%–Add metric
Build a cash flow statement I can switch between quarters and years, add metrics to, and break down by entity.
- Funnel stepsdrag to reorderAdd step1Signed up2002Placed an order1943Bought Software1044Reached 3 orders71Entered200Converted71Reached 3 ordersConversion35.5%window: 1 yearFunnelclick a drop-off to see who left200100.0%1Signed upentry−619497.0%2Placed an order97.0% of prev−9010452.0%3Bought Software53.6% of prev−337135.5%4Reached 3 orders68.3% of prev
Build a funnel tool where I can add, remove and reorder steps and see who converted or dropped off.
- Revenue by region
EMEA is down since the spring promo ended. Check DE first
MTMai TranGive me a freeform canvas where I can place charts anywhere and click a mark to drill into a linked chart.
Bring your favorite agents
Build with any AI agent your team already pays for, like Claude, Cursor or ChatGPT. anfra adds the data and the guardrails, not another AI bill
- Works with any agent that speaks MCP
- No extra model costs to run anfra
- Agent-friendly by design
Claude
Cursor
CodexInspectable by default
Every number in an app traces back to a governed metric, and every step can be opened up and checked, from the chart to the query to the SQL that ran
- Inspect any number: the metric, the query and the SQL behind it
- Filters, drills and breakdowns are inspectable too
- Access rules apply to every query
What people and agents say about anfra
“After about 30 minutes of setup, we had key dashboards and apps with built-in interactions, out of the box, to track our usage.”
GPT 6.1 SolBuilt-in interactions
Apps ship with the interactions people expect from BI, without asking the agent to rebuild them each time
Pick a filter value or click a bar, and every chart on the page follows
Click a bar to go from year to quarter to month
Open the rows behind any number, then export them
Coming soon
Extend your BI with plugins
Turn on the features your team needs in every app, or build your own plugins with your agent
How anfra-built BI is different
Traditional BI is governed but rigid. AI-generated data apps are flexible but ungoverned. anfra-built BI gives you both
| Capability | anfra-built BI | Traditional BI | AI-gen data apps |
|---|---|---|---|
| Trusted numbersOne governed definition of every metric | Yes | Yes | No |
| Access controlRow and column rules follow each viewer | Yes | Yes | No |
| CollaborationPublish, share and work on apps together | Yes | Yes | No |
| Predefined interactionsFilters, drill-downs and exports out of the box | Yes | Yes | No |
| Fast to buildFrom a prompt to a working app in minutes | Yes | No | Yes |
| Customizable and flexibleAny layout or interaction, not just dashboard tiles | Yes | No | Yes |
| AI friendlyAs-code, composable, native to AI | Yes | No | Yes |
Infrastructure that enables governed flexibility
anfra sits between your app and your semantic layer. Apps don’t contain SQL: they ask for metrics by name, and anfra turns each request into a query on the metrics your semantic layer defines.
A JavaScript library your app uses to request data and wire up filters, drill-downs and period comparisons. It asks for data by dataset and metric names, not SQL.
Turns each interaction into a query on your semantic layer, runs it on your warehouse and returns the results. It also serves MCP so coding agents can read your models.
Your models, datasets and metrics, defined as code. anfra isn’t a semantic layer itself, but it ships with AMQL built in, so you can start without one. Support for external semantic layers such as dbt, Cube, Snowflake and Databricks is planned.
Build locally, deploy anywhere
anfra runs on your machine. When your app is ready, deploy it to your own host or to anfra.ai
Install the binary, then the skills that teach your coding agent to build with anfra.
curl -fsSL https://anfra.ai/install.sh | bash then anfra skills install
Add your warehouse credentials to
.anfra/data_sources.ymlanfra init my-project
Open the project in your coding agent and say what you need. It defines any missing metrics as code for you to review.
cd my-project && claude then Use the build-data-app skill. Build a revenue overview with a region filter.
Run
anfra serveand open the app. Inspect any number to see its query and metric.anfra serve
One command puts the app on anfra.ai, where your team can open it
anfra publish
Revenue overview
- Revenue
- $0K +18.2%
- Orders
- 0
- Avg. order
- $60.08
Revenue by region
Monthly trend