Yesod Software Factory

Yesod is my AI software factory. It coordinates the full lifecycle of work—from capturing a feature request or bug report through planning, agent dispatch, implementation, and independent verification. When the deliverable is code, that lifecycle can continue through merge.

The goal is not simply to run more coding agents. It is to build a dependable production system around them: one that gives each agent the right context, keeps work moving through explicit stages, verifies changes before integration, and preserves an auditable record of what happened.

What Yesod Does

Captures work clearly: Feature requests, bug reports, operational knowledge, and implementation history stay connected rather than disappearing into chat transcripts.

Plans and dispatches: Yesod breaks work into actionable units, routes them to appropriate agents, and tracks dependencies between concurrent efforts.

Routes work by capability and cost: Different stages can use different models and agent harnesses, with complexity, capacity, and cost informing the choice.

Verifies before merging: Code changes move through project-specific quality gates and merge lanes so the exact candidate is tested before it reaches the target branch.

Keeps humans in control: The factory automates execution while preserving visibility into decisions, progress, failures, and the evidence behind each result.

Builds institutional memory: Tool knowledge, workflow details, implementation notes, and audit history accumulate into a durable operating system for future work.

Learn More

Visit yesod.work to learn more about the project and how the software factory operates.

I am also developing MeshCrawler's Software Factory as the commercial home for this work.