Architecture

Follow a review through Ironwood.

Ironwood separates the web experience from background review processing so engineers can submit work, follow progress and inspect results without treating the review engine as a black box.

Web appReview request
APIState and orchestration
QueueBackground work
Review workerCode and AI analysis

Core components

Web interface

Where an engineer selects the repository, review scope and provider configuration, then reads the result.

API

Creates the review request, records its state and hands the work to the review queue.

Review queue

Separates the user-facing request from background review processing.

Review worker

Loads the selected code and context, calls the configured AI provider or providers and persists the result.

PostgreSQL

Stores users, configuration, review history, findings and other application records.

Configured providers

Source-control, AI, requirements and delivery systems selected by your organisation.

Where the data lives

The Ironwood application, PostgreSQL database, review queue and review worker run in the environment your organisation controls. Review history, findings, engineering knowledge and configuration are stored with that deployment.

External connections

Ironwood connects only to services your organisation configures, such as GitHub, Azure DevOps, Bitbucket, OpenAI, Claude, Gemini, Ollama, Jira, Azure Boards and Confluence.

The licence boundary

The Ironwood-hosted licence service is separate from the review workload. It validates evaluation and paid licence state. Source code and review content do not travel through it.