atrixANALYTICS

Self-hosting

The licence, what runs where, and the state of the self-host bundle.

Licence

Atrix Analytics is open-core:

PartLicence
SDKs, event contracts, the statistics engine, the CLI and codegenMIT
Control-plane API, console, data plane (ingest, processing, query compiler, region API), MCP server, the basic AI analyst, deployment filesAGPL-3.0
Proactive findings, replay summaries and semantic search, SAML/SCIM, custom roles, audit export, warehouse-native export, multi-region federation, billingCommercial (ee/)

The AGPL lets you run, study, modify and self-host the platform. If you offer a modified version to others over a network, you must make your modified source available to them under the same licence. Nothing outside ee/ depends on ee/, so an AGPL-only build is complete for the core product. The basic AI analyst works self-hosted with your own model key.

What a deployment is made of

ServiceLanguageRole
atrix-apiRustControl plane: accounts, organisations, projects, keys, tokens, OAuth, billing
atrix-captureRustIngest: /e, /r, /wh
atrix-processorRustRedpanda → deduplication → ClickHouse, identity
atrix-flagsRustFlag evaluation and definitions
atrix-region-apiRustQueries, SQL, persons, replays, Ask
atrix-replayRustReplay chunks → object storage
atrix-mcpRustThe MCP server
atrix-erasure, atrix-exportRustGDPR erasure phases; Parquet export to S3
atrix-revenueRustRevenue webhook normalisation and currency conversion
consoleTypeScriptA static single-page app

Backing services: PostgreSQL (a global and a regional database), ClickHouse, Redpanda (Kafka API), Valkey, and S3-compatible object storage (the bundle uses Garage).

Status

In progress

A supported self-host bundle, with Docker Compose for a single machine and a Helm chart for Kubernetes plus published AGPL images, is being prepared for general availability. Today the repository contains the development stack and the production Compose file we run ourselves; they are the reference, not yet a supported product.

To try the whole stack locally from a checkout (Docker, the pinned Rust toolchain and Bun required):

bun install
bun run dev:all        # the docker stack, migrations, every service and the console; Ctrl-C stops it

The console is then on http://localhost:4500. bun run sim:load loads a synthetic dataset if you want data to explore.

On this page