Architecture

Django Automate is built on a modular "Engine + Plugins" architecture.

graph TD
    API[API Gateway] --> Dispatcher
    Webhook[Webhook Ingress] --> Dispatcher

    Dispatcher --> Queue[(Job Queue)]

    Queue --> Worker[Worker Pool]

    Worker --> Engine[Execution Engine]

    Engine --> Provider[Provider Registry]
    Engine --> Store[Vector Store]
    Engine --> DB[(Postgres)]

    Provider --> LLM[LLM APIs]
    Provider --> Tools[External Tools]

Core Components

1. Job Queue (The Spine)

All work is asynchronous. Job is the canonical unit of work. Backends (Celery, Redis, SQS) are just transports.

2. Execution Engine

A directed acyclic graph (DAG) executor that steps through Workflow definitions. It handles state persistence, retries, and error handling.

3. Provider Registry

A unified interface for external capabilities. Whether it's OpenAI, a local vector store, or a Slack connector, they all implement BaseProvider.

4. Admin Studio

The control plane. It provides visibility into queues, executions, and allows for "Human in the Loop" interventions.