intraQ — Self-hosted AI BI with a Knowledge Layer

Website · Docs · Quickstart · Configuration · Comparisons · Discussions · Contributing

intraQ is a source-available AI business intelligence platform that lets teams query SQL-backed operational data using natural language, generate trusted SQL, and build interactive dashboards without traditional BI complexity.

It combines AI analytics, natural language SQL, text-to-SQL workflows, an AI dashboard builder, embedded analytics foundations, MCP tools, RAG-style context retrieval, and a semantic layer built from local metadata, data dictionary entries, SQL models, relationships, dashboard context, and safe result summaries.

Why intraQ?

Most AI reporting tools stop at “generate a chart” or “write a query.” intraQ is built around verified operational BI:

• Answers stay tied to evidence — SQL, selected data source, result rows, and assumptions remain inspectable.
• Dashboards come from real queries — AI-created components are saved as live dashboard elements, not static screenshots.
• Business context is explicit — dictionaries, SQL models, relationships, metrics, and dashboard context guide the AI before it answers.
• Unsafe gaps are visible — when data, filters, or model context are missing, intraQ says what is missing instead of inventing an answer.
• Operational use cases come first — revenue health, product mix, wastage signals, account risk, billing exceptions, and embedded customer reporting.

The goal is not just natural language SQL. The goal is AI BI that a team can inspect, reuse, and improve.

Why teams use intraQ

• AI business intelligence — ask operational questions in plain English.
• Trusted SQL generation — inspect SQL and evidence before publishing results.
• AI dashboard builder — turn answers into reusable dashboards.
• Self-hosted BI — run with Docker Compose against your own database.
• Semantic layer — define tables, fields, joins, metrics, and business meaning.
• LLM-ready context — ground answers in local metadata and RAG-style knowledge retrieval.
• Embedded analytics foundation — use intraQ as a reporting layer for product and internal workflows.
• Provider-flexible AI — configure Codex OAuth, OpenAI, or Gemini from the admin UI.

Who is this for?

intraQ works with any industry — if your operational data lives in a SQL database, intraQ adapts to it. There is no vertical lock-in; you define what your data means and intraQ grounds its answers in those definitions.

It's built for teams that want answers from operational data without a dedicated BI function:

• Operations & business teams — ask questions in plain English and get trustworthy answers, no SQL or analyst required.
• Platform & SaaS vendors — embed white-label AI analytics for your own customers instead of building it in-house.
• Internal BI & data teams — self-service analytics grounded in governed, inspectable definitions.

Connect any Postgres, MySQL, or MSSQL database and start asking questions.

Want a tailored, pre-built knowledge base and hands-on support for your domain? That's available commercially — see intraq.dev.

Demo flow

Ask a business question
        ↓
AI plans against metadata and SQL models
        ↓
Trusted SQL is generated and executed
        ↓
The answer returns with evidence
        ↓
Save the result as a live dashboard component

Use the seeded demo to try questions such as:

• How is revenue trending by day?
• Which channel has the highest revenue?
• Compare revenue and gross margin by category.
• Which location has the highest average order value?
• Create a dashboard chart for revenue by channel.

See the full Demo guide.

Architecture

User question
  → AI Analyzer
  → Metadata + semantic model + saved SQL models
  → SQL planning and validation
  → Connected operational database
  → Result summary + evidence
  → Dashboard Builder
  → Saved live dashboard

The public source includes local dashboards, Analyzer, SQL models, MCP, data-source management, and dashboard-builder workflows.

It intentionally excludes paid AI Studio, proprietary domain intelligence, control plane, paid release tooling, private operational docs, generated artifacts, credentials, and private operational material.

Use cases

intraQ is built for operational reporting where users need more than static charts — across any industry with SQL-backed data:

• Revenue & sales analytics — trends, margins, channels, products, and location performance.
• Operational & performance reporting — throughput, exceptions, and period-over-period comparisons.
• Embedded customer-facing analytics — dashboards over product data inside your own application.
• Financial & risk reporting — accounts, balances, aging, exposure, and payment behaviour.

You define the metrics and relationships for your domain; intraQ grounds its answers in them. Explore practical question sets in examples/.

Comparisons

intraQ is not trying to replace every enterprise reporting suite. It is focused on operational BI workflows where teams want AI-assisted SQL, local control, and dashboards from governed data models.

Capability	intraQ	Power BI	Metabase	Lightdash
Plain-English → SQL	✅	Partial	Partial	Partial
AI-built dashboards	✅	Partial	❌	❌
Semantic / knowledge layer	✅	Partial	Limited	✅ (dbt)
Evidence & SQL shown for every answer	✅	❌	Partial	Partial
Self-hosted	✅	❌	✅	✅
Embeddable / white-label	✅	Limited	Limited	Limited
No data team required	✅	❌	Partial	❌

Detailed comparison pages:

• intraQ vs Metabase
• intraQ vs Wren AI
• intraQ vs Power BI
• intraQ vs Lightdash

intraQ vs Power BI

Power BI is a broad reporting suite. intraQ focuses on natural language SQL, evidence-backed AI answers, and turning operational questions into live dashboard components from SQL-backed models.

intraQ vs Metabase

Metabase is strong for self-service querying and dashboards. intraQ adds an AI Analyzer and dashboard-builder workflow designed around plain-English questions, trusted SQL generation, and evidence review.

intraQ vs Looker

Looker centers on governed semantic modeling at enterprise scale. intraQ is lighter to run locally and focuses on operational teams that want AI-assisted reporting over existing SQL data.

intraQ vs Sigma Computing

Sigma provides spreadsheet-style cloud analytics. intraQ focuses on self-hosted or controlled deployments, natural language SQL, and operational dashboards grounded in local metadata and models.

Quickstart

The shortest path is Docker Compose:

cp .env.example .env
docker compose up --build

Then open http://localhost:4100.

Seeded local login:

Email	Password
[email protected]	intraq-demo

For local development without Docker:

nvm use
npm ci
cp .env.example .env

Edit .env and set:

DATABASE_URL=postgresql://user:password@localhost:5432/intraq
AUTH_TOKEN_SECRET=replace-with-at-least-32-random-characters

Then run:

npm run db:migrate
npm run db:seed
npm run dev

More setup detail is in QUICKSTART.md. Environment variables are documented in docs/CONFIGURATION.md.

AI provider setup for Codex OAuth, OpenAI, and Gemini is documented in docs/AI_PROVIDER_SETUP.md. For Codex OAuth browser login, set OPENAI_OAUTH_CLIENT_ID in the API environment; otherwise the admin page cannot start the Codex login flow.

Development

npm test
npm run build

Copy .env.example to .env for local development. Do not commit local env files, database passwords, provider keys, client data, or private operational material.

Focused workflow docs:

• AI Analyzer
• Dashboard Builder
• SQL Editor
• MCP tools
• AI provider setup
• Comparisons
• Publication checklist

License and public source scope

intraQ is source-available under the IntraQ Sustainable Use License.

You may use, fork, modify, and run intraQ for internal business, personal, educational, evaluation, and non-commercial purposes.

Paid hosting, managed service use, white-label resale, OEM redistribution, paid third-party support/operations, or use in a competing commercial analytics, BI, dashboard, SQL-assistant, or AI-reporting service requires a commercial agreement with IntraQ.

See LICENSE.md, COMMERCIAL.md, docs/PUBLIC_SOURCE_SCOPE.md, CONTRIBUTING.md, and SUPPORT.md.

Suggested GitHub topics

Recommended repository topics:

ai-analytics, ai-bi, llm, rag, semantic-model, semantic-layer, business-intelligence, analytics-platform, ai-dashboard, ai-dashboard-builder, text-to-sql, dashboard-builder, self-hosted, postgres, sql, embedded-analytics, reporting, natural-language-sql, natural-language-query, operational-analytics

Use agentic-ai only if you want the repository positioned around autonomous multi-step AI planning. If the public source is presented mainly as governed AI reporting and dashboard generation, the topics above are more precise.