Skills

• SkillUSD 2

  Prompt vs RAG vs Fine-tune

  Decides how an AI capability should be built — better prompting, retrieval over your data, fine-tuning, or a combination — from the actual requirement, with the costs and failure modes of each path stated before money gets spent on the wrong one.

  Best for: The 'should we fine-tune?' meeting that recurs quarterly without a framework

• SkillUSD 2

  User Interview Script

  Writes the interview guide that gets you truth instead of politeness: questions about past behaviour rather than imagined futures, probes planned in advance, and the leading questions removed before they poison the data.

  Best for: Founders about to interview users for the first time

• SkillUSD 2

  Survey Design

  Designs surveys whose results you can actually use: questions that measure one thing each, scales that mean something, the biases engineered out — and the analysis planned before the first response arrives.

  Best for: Surveys that return 'interesting' data nobody can act on

• SkillUSD 2

  Persona Draft

  Builds personas from evidence instead of imagination: the behavioural patterns your research actually shows, the goals and constraints that drive decisions — and none of the stock-photo fiction that makes personas deservedly mocked.

  Best for: Teams whose personas are demographic fiction nobody consults

• SkillUSD 2

  Journey Map

  Maps the customer's actual journey through your product or service: the steps as they experience them, the emotions and friction at each, the backstage systems involved — exposing where the journey breaks and where effort is wasted on steps that already work.

  Best for: Teams optimising their own silo while the handoffs bleed customers

• SkillUSD 2

  Feature Brief

  Writes the one-page brief that aligns everyone before a feature gets built: the problem with its evidence, who it serves, what success measurably means, what is out of scope — the thinking, forced onto paper, where disagreement is cheap.

  Best for: Features that start building before anyone agrees what they are

• SkillUSD 2

  Prioritisation Pass

  Runs a structured prioritisation over your backlog: value and effort estimated honestly, the scoring model fitted to your strategy, sacred cows examined like everything else — producing an order you can defend and a cut line you can say out loud.

  Best for: Backlogs ordered by whoever asked loudest or latest

• SkillUSD 2

  Roadmap Draft

  Drafts a roadmap that communicates without lying: themes and outcomes on honest horizons — now, next, later — instead of feature-date promises that decay into apologies, fitted to the audience reading it.

  Best for: Roadmaps that are date-commitments nobody believes by month two

• SkillUSD 2

  Stakeholder Update

  Writes the recurring update that keeps stakeholders informed and off your back: progress against what was promised, the problems stated before they are discovered, the asks made explicit — in the five minutes they will actually give it.

  Best for: Updates that are either novels nobody reads or vagueness nobody trusts

• SkillUSD 2

  Launch Checklist

  Builds the checklist that makes a launch boring: everything that must be true before, during and after — across product, comms, support and measurement — with owners and go/no-go criteria, so launch day executes instead of improvises.

  Best for: Launches that discover their gaps live ('did anyone tell support?')

Also see the Agents.