Skill Library

Generate acceptance criteria for AI features with non-deterministic outputs. Define what 'done' looks like including quality thresholds and failure categories.

Eval · Quality · Drift

Design an AI agent from user goal to autonomous execution. Define tools, autonomy levels, and how failures are handled.

Autonomy · Tools · Memory

Run A/B tests on AI features correctly. Design experiments for AI-powered features and interpret results despite non-determinism.

Eval · Quality · Drift

Define what AI-assisted code review should and shouldn't catch. Set up team standards on appropriate use of AI code review tools.

Build · Ship · Scale

Build an AI governance framework. Define processes, ownership, and standards that govern how AI features are built, reviewed, and monitored.

Safety · Risk · Policy

Spec AI memory systems for personalisation and agents. Design how an AI product remembers across sessions and agent runs.

Autonomy · Tools · Memory

Identify and build defensible AI moats. Assess competitive defensibility and plan investments to widen the gap.

Vision · Discovery · Moat

Design AI-powered onboarding that activates users faster. Personalise, accelerate, or improve the new user experience.

Experience · Trust · UX

Write specs and briefs that work as effective AI pair-programming inputs for engineers using Cursor, Copilot, etc.

Build · Ship · Scale

Diagnose and strengthen AI product-market fit. Assess whether your AI feature has found PMF and identify blockers.

Vision · Discovery · Moat

Define AI quality metrics that your team can actually track in production. Output quality, system reliability, and user impact.

Eval · Quality · Drift

Run a structured red-team exercise on an AI feature to find failure modes before users do. Identify exploits and harmful outputs.

Eval · Quality · Drift

Define AI safety requirements. Identify safety controls that prevent dangerous, harmful, or out-of-control outputs.

Safety · Risk · Policy

Build an AI-driven experimentation system that runs, analyses, and learns from product experiments at scale.

Vision · Discovery · Moat

Specify an automated QA pipeline for AI features and AI-generated code. Define test strategy for LLM and agent features.

Build · Ship · Scale

Define guardrails for autonomous AI task execution. Safety limits, circuit breakers, and human oversight mechanisms.

Autonomy · Tools · Memory

Evaluate whether to build, buy, or partner for an AI capability. Structured recommendation with scorecard and TCO.

Vision · Discovery · Moat

Set up an AI-assisted competitor monitoring pipeline. Track activity, analyse implications, build a weekly system.

Vision · Discovery · Moat

Set up confidence scoring so users know when to trust the AI and when to verify. UX, calibration, and signals.

Experience · Trust · UX

Manage context windows for complex PM tasks involving large documents, long conversations, or multi-session work.

Experience · Trust · UX

Build and run a continuous discovery system. Always-on intake, automated tagging, weekly synthesis, opportunity backlog.

Vision · Discovery · Moat

Decide between copilot and agent interaction models. Choose the right autonomy tier with a structured framework.

Autonomy · Tools · Memory

Design a data flywheel that makes your AI product harder to copy over time. Convert usage into model improvements.

Vision · Discovery · Moat

Choose and spec embedding models for semantic search, RAG, recommendations, or similarity tasks.

Build · Ship · Scale

Build an evaluation framework for an LLM-powered feature before it ships. Functional, quality, adversarial, regression.

Eval · Quality · Drift

Specify explainability requirements for AI decisions. Output, decision, and audit-level explanations.

Safety · Risk · Policy

Evaluate and reduce hallucination risk in AI features. Safeguards proportional to stakes plus monitoring.

Safety · Risk · Policy

Design human-in-the-loop evaluation workflows. Who reviews, what they look for, how findings feed back.

Eval · Quality · Drift

Spec hyper-personalisation features driven by AI. Adapt content, recommendations, or experiences to individuals.

Experience · Trust · UX

Evaluate and specify MCP (Model Context Protocol) integrations. Build vs use existing servers, security and trust.

Autonomy · Tools · Memory

Review and write model cards. Evaluate third-party cards or document internally-built models with accountability.

Safety · Risk · Policy

Detect and respond to model drift before users notice. Build a drift detection and response system.

Eval · Quality · Drift

Specify requirements for multi-agent systems. Orchestration, handoff protocols, contracts, and failure handling.

Autonomy · Tools · Memory

Decide between on-device and cloud AI inference. Privacy, latency, cost, capability, and offline tradeoffs.

Build · Ship · Scale

Stress-test a PRD or AI feature spec. Four critic perspectives surface every gap before the team sees it.

Vision · Discovery · Moat

Use AI to identify and act on churn signals before users leave. Predict, intervene, measure.

Vision · Discovery · Moat

Design and build prompt chains for multi-step PM tasks where a single prompt isn't enough.

Build · Ship · Scale

Write a production-quality system prompt and hand it off as a first-class spec artefact for engineering.

Build · Ship · Scale

Choose the right prompting approach: zero-shot, few-shot, chain-of-thought, or RAG. Tradeoffs and decision tree.

Build · Ship · Scale

Synthesise qualitative research from multiple sources into structured insights without losing nuance.

Vision · Discovery · Moat

Build responsible AI requirements into a product from day one. Five pillars: fairness, transparency, privacy, safety, accountability.

Safety · Risk · Policy

Design semantic search. Query pipeline, indexing, ranking logic, and quality metrics.

Build · Ship · Scale

Detect and report spec drift between an original AI feature spec and what was built.

Vision · Discovery · Moat

Turn a completed feature spec into an engineering-ready handoff package with everything engineering needs.

Vision · Discovery · Moat

Write AI feature specs using GIVEN/WHEN/THEN format. Turn a rough idea into a structured spec.

Vision · Discovery · Moat

Build synthetic persona sets to pressure-test product assumptions when real user access is limited.

Experience · Trust · UX

Write effective system prompts that reliably shape AI behaviour. Five-layer prompt architecture.

Build · Ship · Scale

Design and run AI-assisted user interviews. Pre-interview, post-interview, and cross-interview synthesis.

Vision · Discovery · Moat

Evaluate and specify vector database requirements. Managed, self-hosted, pgvector, in-memory comparison.

Build · Ship · Scale

Use AI coding tools effectively as a non-engineering PM to prototype, explore, and communicate technical ideas.

Experience · Trust · UX