Introduction
What is Copula
Copula is a multi-agent AI development team. A task travels through a pipeline of specialized LLM agents, each owning one role:
- Product — turns a business request into a BRD.
- RnD — produces the technical specification (V&S).
- Planner / JobDecomposer — decomposes the spec into executable jobs.
- Developer — writes the code.
- QA — verifies the result and decides pass/fail.
- DevOps — deploys.
- E2E — exercises the running UI.
- Chronicler — keeps the log.
The core pipeline is Task → Developer → QA → Delivery (→ E2E): a task is implemented by the Developer agent, checked by QA, delivered, and optionally verified end-to-end against the running application.
The stack: a Go backend, a Vue 3 frontend, PostgreSQL, Docker, and any OpenAI-compatible LLM provider (including Copula's own cabinet LLM router).
Cloud vs self-hosted
Both editions are the same product; they differ in who runs the infrastructure.
| Cloud | Self-hosted | |
|---|---|---|
| Infrastructure | Managed by the Copula team | Runs entirely on your own server via Docker |
| Setup | Sign up in the web cabinet and start working | Provision a server (wizard or manual compose) |
| Access control | Cabinet account | A license token issued in the cabinet |
| Data | Copula infrastructure | Stays on your server |
Self-hosted Copula runs the full stack on your machine under a license token — see the Self-hosting section for the complete guide.
Where to go next
- Self-hosting overview — deploy Copula on your own server.
- Requirements — what the target server needs.
- Getting a license — the license token from the Copula cabinet.
- Guides — working with the agent pipeline (coming soon).
- Reference — configuration reference (coming soon).