Skip to content
SYNAPLAN

DEPLOY // SYS-04

Self-hosted AI — Docker, Kubernetes or Germany.

  • Self-hosted AI
  • On-premise AI
  • Docker Compose
  • Kubernetes
  • Helm chart
  • EU hosting
  • openDesk

The open-source AI platform ships as Docker images and a public Helm chart. On-premise on a laptop or a server, Kubernetes in your cluster, or EU hosting in Germany. Same Apache-2.0 code. No enterprise fork.

laptop · docker compose
$ git clone https://github.com/metadist/synaplan.git
$ cd synaplan
$ docker compose up -d

# optional: local models on your own hardware
$ COMPOSE_PROFILES=local-ai docker compose up -d

→ http://localhost:5173
01 /Laptop

docker compose up -d

The complete development stack — backend, worker, frontend, MariaDB, Redis, Qdrant, Ollama — in about ten minutes on any machine with Docker.

  • ~3 GB standard install, cloud AI
  • + local AI via the local-ai compose profile
  • Linux, macOS (Apple Silicon native), Windows/WSL2
laptop · docker compose
$ git clone https://github.com/metadist/synaplan.git
$ cd synaplan
$ docker compose up -d

# optional: local models on your own hardware
$ COMPOSE_PROFILES=local-ai docker compose up -d

→ http://localhost:5173
02 /Server

The deploy/ contract

One Linux server, pre-built images from ghcr.io, a reverse proxy in front. Migrations run on container start. Documented as a contract, not a wiki page.

  • ghcr.io/metadist/synaplan
  • Automatic TLS via Caddy
  • Runs on 4 vCPU / 8 GB

T-02

The deploy/ contract

  • ghcr.io/metadist/synaplan
  • Automatic TLS via Caddy
  • Runs on 4 vCPU / 8 GB
03 /Cluster

helm install synaplan

Public Helm chart for Kubernetes: multiple web replicas, external MariaDB Galera, Qdrant, Redis, GPU workers for local inference. Built for hosting providers.

  • oci://ghcr.io/metadist/synaplan-charts
  • Galera-safe migrations
  • Optional Triton / Ollama GPU nodes
cluster · kubernetes / helm
$ helm install synaplan \
    oci://ghcr.io/metadist/synaplan-charts/synaplan \
    --namespace synaplan --create-namespace \
    -f values.yaml

# GPU inference sidecar (optional)
$ helm install triton oci://ghcr.io/metadist/synaplan-charts/triton

→ deployment/synaplan-web  3/3 ready
04 /Sovereign cloud

openDesk and EU hosting

Synaplan integrates with openDesk and runs on European infrastructure. Or skip the ops entirely and use our platform hosted in Germany.

  • openDesk installer
  • web.synaplan.com — hosted in Germany
  • synaplan.eu for public administration

T-04

openDesk and EU hosting

  • openDesk installer
  • web.synaplan.com — hosted in Germany
  • synaplan.eu for public administration
05 /Minimum requirements

  • Runtime

    Docker 24+ with Compose v2, or Kubernetes 1.28+

  • Database

    MariaDB 11.7+ (VECTOR) — bundled or external Galera

  • Vectors

    Qdrant (bundled) or MariaDB VECTOR

  • Models

    Ollama locally, or any supported cloud API key

T-00:00:10

The docs are the manual. Not marketing.

Getting started, production deployment, Kubernetes, feature modules — all on docs.synaplan.com.

[ Open deployment docs ]Helm charts on GitHub ↗