Aito — the predictive database

Free for development and CI. Licensed for production.

docker pull ghcr.io/aitohq/aito
docker run -p 9005:9005 ghcr.io/aitohq/aito

…or pull from the AWS mirror:

docker pull public.ecr.aws/aitoai/aito

Quickstart

docker run -d \
  --name aito \
  -p 9005:9005 \
  -v aito-state:/io/state \
  ghcr.io/aitohq/aito:latest

# Insert a row
curl -X POST http://localhost:9005/api/v1/data/companies \
  -H 'content-type: application/json' \
  -d '{"name":"acme","revenue":1000000}'

# Predict
curl -X POST http://localhost:9005/api/v1/_predict \
  -H 'content-type: application/json' \
  -d '{"from":"companies","predict":"revenue"}'

Full docs: https://aito.ai/docs

Free-tier limits

The default image is free for development and CI:

Limit	Default
Rows per table	10,000
Rows total	50,000

Going over either limit returns HTTP 429 with {"error":"row_limit_exceeded"}. The server keeps serving queries against existing data; only inserts past the limit are rejected.

To remove the limits, set AITO_LICENSE_KEY to a key issued by console.aito.ai:

docker run -d \
  -e AITO_LICENSE_KEY=ak_live_… \
  -p 9005:9005 \
  ghcr.io/aitohq/aito:latest

The image phones home to console.aito.ai/public/licenses/validate on startup. The response is cached (AES-encrypted on the /io/state volume) for up to 7 days, so the image keeps working through network outages.

Configuration

Variable	Default	Meaning
PORT	9005	HTTP listen port
BIND_ADDRESS	0.0.0.0	Listen address
STATE_PATH	/io/state	Where the database persists
JVM_XMX	2g	JVM max heap
JVM_XMS	512m	JVM initial heap
AITO_LICENSE_KEY	(unset)	License key for production use
AITO_LICENSE_API	https://console.aito.ai	License validation endpoint
AITO_LICENSE_CACHE_FRESH_SECONDS	86400 (24h)	Skip-network window
AITO_LICENSE_CACHE_MAX_AGE_SECONDS	604800 (7d)	Hard cache TTL

How this image is built

This repo doesn't hold any of the engine code. On a v* tag push (or workflow_dispatch), .github/workflows/publish.yml:

1. Downloads the obfuscated free-tier JAR from the matching AitoDotAI/aito-core release.
2. Builds a thin Alpine + JRE 17 image around it.
3. Smoke-tests the image boots and answers /status.
4. Pushes to ghcr.io/aitohq/aito:<tag> and public.ecr.aws/aitoai/aito:<tag> (+ :latest).

The engine lives at AitoDotAI/aito-core. Bugs and feature requests there.

License

The Docker image is freely distributable under the terms documented at https://aito.ai/license. The contents of this repo (Dockerfile, entrypoint, publish workflow) are MIT-licensed.