Evaluation
Docker Compose
One file starts Postgres, Kafka, Debezium, Elasticsearch, and the ArtificialFlow services, and connects them to your OIDC provider. This path is for a laptop or a lab, not a production cluster.
Generate ComposeOpen source · v1.1.0
BPMN workflows on infrastructure you run.
ArtificialFlow is a self-hosted platform for modeling, executing, and observing business processes. A browser modeler, a Go engine, and a Node.js SDK share one gateway.
# evaluation, published images
docker compose -f artificialflow.compose.yml up -d
# production chart
helm upgrade --install artificialflow ./charts/artificialflow \
--namespace artificialflow --create-namespace \
-f ./artificialflow-values.yaml
UI at http://localhost:9100
What the platform includes
The command database is the source of truth. Search-backed views are eventually consistent.
Model
Browser BPMN modeler, process catalog, and versioned deployments.
Execute
Timers, messages, signals, decision tables, external workers, and idempotency.
Observe
Dashboards, instance history, incidents, and a read-optimized query API.
Integrate
OIDC for people. A Node.js SDK and HTTP API for workers and automation.
How a request moves
Browsers and the SDK enter through the gateway. Writes land in Postgres. The sync worker projects them into search for the query API.
01
Gateway
One public entry. Routes the UI, command API, and query API.
02
Command
Deploys processes, mutates instances, and writes the source of truth to Postgres.
03
Runtime
Advances tokens, timers, jobs, and SLA checks.
04
Sync
Projects outbox events and Debezium changes into search.
05
Query
Serves process and instance reads from Elasticsearch.
06
IAM
Validates people and workers through your OIDC provider.
Production
Helm on Kubernetes
The chart expects Postgres, Elasticsearch, and Kafka with Debezium that you already operate. The builder writes an environment values file and the install command for chart v1.1.0.
Generate Helm values