Frameworks

Reusable Product OS systems. Start with the problem, then the layer coverage, then the implementation artifacts.

Product OS model

Four-layer operating model

A single public framework language for how product systems run in an AI-driven execution environment.

Layers: Product (strategy, customer, hypotheses, measurement), Execution (requirements, context, agents, implementation, validation), Memory (state, decisions, constraints, learnings), Human (judgment, approval, escalation, strategy).

Evidence: Labs case studies · Ideas: Thinking

See experiments

Problem-first framework

Angry Letters operating system

For solo execution of a live SaaS where speed matters but accountability and context cannot break.

Layer coverage: Strong in Execution and Human layers with practical Memory structures for campaign operations, approvals, and delivery constraints.

Evidence: Angry Letters case study · Ideas: Thinking

Repository (Clone / Download)

Problem-first framework

Product strategy and context system

For teams that need durable strategy context across messy inputs, long timelines, and multi-agent execution cycles.

Layer coverage: Strong in Product and Memory layers with explicit handoffs into Execution and Human approval loops.

Evidence: matthewheyman.com case study · Ideas: Thinking

Repository (Clone / Download)

Legacy technical identifiers: Some repositories and install paths still include historical names such as matthew-heyman-oais, angry-letters-oais, poais-core, and oais/. These are retained for compatibility while public category language stays Product OS.