Kartini Cooper

The first job of any transformation is honest diagnosis. Six dimensions, six stages, scored by a cohort, calibrated openly. The output is a maturity baseline the rest of the methodology runs on — and the place where the gap between perceived and actual readiness usually surfaces. Halcyon Financial's pre-programme assessment is the worked example.

Turning the diagnosis into three OKRs the organisation can commit to honestly. Calibrated confidence, premortems, backcasted key results. Borrows openly from Annie Duke's Thinking in Bets. The discipline isn't picking the right targets — it's setting them as bets, not promises. Halcyon's annual OKR slate is the worked example, with the Governance & Risk KR illustrating how to measure honestly when single instruments fail.

Improvement ideas surface in every retro. Most die from time pressure and the loudest voice. A lightweight intake, an effort-impact rubric, a triage cadence, and a single-page RACI — the operating model that turns scattered ideas into a prioritised portfolio. Halcyon Financial worked example throughout.

Turning a strategic ambition — "be AI-Native within twelve months" — into three outcomes a sponsor can defend and a sceptic can falsify. Frame, decompose, scope, operate. Where most AI programmes stall is the gap between ambition and a tractable workstream. This is the bridge.

Twelve vendors longlisted, three shortlisted, one chosen — without losing optionality. Frame, shortlist, evaluate, onboard. The discipline is to choose for the exit, not the entry. Halcyon's selection of Lighthouse AI as the delivery partner for the CX tool walks the full method.

The six weeks of change-readiness work that surrounds an AI launch — without making promises the programme can't keep. Listen, frame, equip, measure. Trust as the leading indicator, not the lagging one. The most workforce-sensitive of the six. Maya Chen is the human anchor.

Most "AI doesn't work for me" is a prompting problem, not a model problem. Six patterns, six templates, six before/afters — the brief, the exemplar, the critic, the structured output, the system prompt, and the chain. The foundation everything else in the series sits on.

Most people use Claude as if every chat were the first. Four layers exist for a reason — chat, Style, Project, Memory. Set them up once and watch a 200-word prompt shrink to 20 with the same output. The leverage layer of the series.

A plain-language, step-by-step guide for non-technical professionals. Told through one real use case: turning messy meeting transcripts into clean action items, the same way every time, in seconds rather than hours.

MCP is the universal adapter for your work. This is the 30-minute walkthrough — using the Friday wrap-up as the worked example. The small ritual that lets the weekend be a weekend.

Define, collect, score, iterate. A handful of real test cases, a clear rubric, a re-run every time something changes. Half a day to set up — and the cheapest piece of infrastructure you'll own. Trust is a record, not optimism.

You've found something that works. Now your team needs to use it too. Find your five, make it visible, lower the floor, make it routine. Bottom-up, not big-bang. The discipline of spreading capability without becoming the help desk.

An "agent" is a thing Claude does — not a thing you build. A Skill, a Project, a few Connectors, arranged so Claude can pull context, draft work, and surface it for you to decide on. The agent does the typing. You do the judging.

One agent is a helper. Four agents passing work between each other is a workflow. The split · the handover · the conductor · the view. Foundations that apply to every team you'll build after this one. The capstone of the series.