Capstone — ship a real deliverable

Everything so far has been practice. The capstone is the performance: one complete, professional deliverable you build with AI as your assistant and defend as your own work. It is the difference between finishing a course and being able to say — and prove — that you can do AI-powered economic analysis to a professional standard.

Choose one of three

Pick the deliverable closest to the work you actually want to do. Each is a genuine professional artefact, not an exercise.

Option 1 — Economic research report

• A focused research question on an African economy (e.g. 'Has mobile-money adoption raised household savings in Kenya?').
• A short literature review (AI-assisted discovery and summarising — every citation verified against the real paper).
• A data section: real data, cleaned and analysed, with at least one chart and one model or descriptive result you can defend.
• Findings and a clear, honest statement of limitations.
• Length: roughly 8–12 pages. Structure: question → evidence → method → findings → limitations.

Option 2 — Policy brief

• A live policy question for a named decision-maker (a county government, the National Treasury, a regulator).
• The standard four-part spine: problem → evidence → options → recommendation.
• Every statistic grounded in a real source (KNBS, CBK, World Bank, a specific budget document) and cited.
• A stakeholder read: who wins, who loses, who can block it.
• Length: 2–4 pages. It must be decision-ready — a busy principal secretary should be able to act from it.

Option 3 — Economic dashboard

• A real dataset (macro indicators, county data, sector data) turned into a small interactive or well-designed static dashboard.
• At least three genuine insights the visuals actually support — not decoration.
• A short written read-out interpreting what the dashboard shows and what it does not.
• Built with AI as a coding and design assistant; you must understand and be able to explain every calculation behind every number shown.

How to use AI in the capstone (and how not to)

Use AI exactly as the course taught: to triage literature, draft prose, scaffold and debug code, summarise long documents, and pressure-test your reasoning. Then do the part AI cannot: check every claim, run the numbers yourself, and take responsibility for the conclusions. A capstone that is visibly an unverified AI dump — confident, fluent, and wrong in three places — is the failure case this whole course exists to prevent.

The rubric — how it is graded

• Rigour (30%): is the analysis correct, appropriately scoped, and honest about its limits?
• Verification (25%): does every figure, quote, and citation trace to a real, checked source?
• Use of AI (15%): was AI used well — as an amplifier — and disclosed?
• Clarity (20%): is it structured, readable, and decision- or reader-ready?
• Relevance (10%): does it engage a real African economic question with real data?

The certificate

When your deliverable is ready, sit the short AI for Economists assessment. Passing it — together with a capstone that meets the rubric — earns a LeadAfrik certificate with a unique ID that anyone can verify online. It is a claim you can defend, because you did the work and checked it yourself.