0:00

KayKav Academy · Domain Expert Builders · Partner Pitch

Built by the people who know

Miracle Agada & Osamudiamen (Mudia) Imasuen — co-founders.
A proposal for governments, foundations and development partners.

Scroll

In summary

Africa has trained millions in tech and digital literacy, yet struggles to find builders where needed. A decade of certificate-based training has yielded credentials over capability. KayKav Academy has solved for this and demonstrated, twice, that non-engineers can deploy real software solutions in four weeks by thinking like product owners and directing AI agents as the build layer.

01The problem

Nigeria and the wider continent have funded technical training for the better part of a decade

National programmes, donor-backed schemes and private academies mostly run on the same model, and the outcome is predictable.

The gap is not intelligence or motivation. Traditional training teaches people to describe software, not to make the decisions that produce it.

Video lessons

Watched, not practised.

Quizzes

Recall, not judgement.

Self-paced platform

No cohort, no deadline, no ship date.

Certificate

A credential employers have learned to discount.

People finish. They have a credential. They still cannot build anything, and employers know it.

02What changed

The cost of building has collapsed. The cost of judgement has not.

Collapsed

The first draft of a build

AI coding agents now produce working software from a clear specification. The typing is no longer the scarce part. Anyone who can direct the work can get a first version in days.

Unchanged, and teachable

Judgement

Framing the right problem. Knowing what shape the solution should take. Directing the build. Evaluating whether it actually works. Keeping it alive afterwards.

Teachable in weeks — and it sits naturally with people who already carry domain expertise.

The person closest to a problem is now the cheapest person to solve it. Domain expertise is the anchor. Everything else is an add-on, and the add-ons have become small.

03The insight

The people who know the problems are not the people building the tools

1
A Public Servant watches a citizen request pass through six desks and knows exactly where it stalls.
2
A laboratory scientist sees sample readings drift between technicians and between shifts.
3
A relationship manager rebuilds the same credit memo every week and knows it is inconsistent.

04Proof · built, not prompted

What we have already proven

0Students across the cohorts

0Countries, 3 continents

0Students learning to solve problems with AI

0Live products across the cohorts

Across the cohorts: 150 students · 7 countries · 3 continents · 70+ live products, and 1000+ students learning to solve problems with AI.

Selected products · every one is on a real URL today

Precisiontryprecisioncare.com Care coordination product A health workflow tool shipped by a non-engineer in four weeks DraftDeskdraftdesk.online Drafting and document workspace The document-heavy pattern most public and legal work runs on Wandarwandar.co Travel itinerary planning Plan, share and remix itineraries with recommendations from real travellers

DDTDon’t Destroy the Structure End-to-end management of non-destructive testing Built by a lab’s own product designer; his employer subscribed for a year

Oneflutetheoneflute.vercel.app AI-assisted PR and creator marketing platform Brands, PR teams and creators in one place: drafting, media outlets, placements and paid bookings

05The programme

Domain Expert Builders: one anchor, five add-ons

The anchor

Years inside a domain. It cannot be bought, and it is what every participant walks in with.

Frame

Turn “this is annoying” into a bounded, measurable system with an accountable owner.

Data

Know where the data lives, what shape it is in, and what you are allowed to touch.

Architecture

Decide whether the answer is a form, a workflow, a dashboard or an app.

Direct

Write specs an agent can build from, then evaluate whether it did what you meant.

Hold

Deploy, secure, run a feedback loop, and hand the tool to someone who keeps it alive.

None of the five is “learn to code”.

06How it runs

Six weeks, from a problem you live with to a tool in use

Week 0

Domain Audit

Inventory ten problems from the working week. Rank by pain and by data reach. Name who is accountable.

Weeks 1–2

Frame & architect

The chosen problem becomes a specification, a data model and a permission design before any code.

Week 3

Build

Directed by the participant, executed by AI agents, reviewed one step at a time.

Week 4

Evaluate & sweep

Five real colleagues use the tool. What failed and what was unnecessary is removed.

Week 5

Ship & hand over

Live deployment, security review, Guardrail sign-off, Handover Pack, named owner.

Week 6

Demo Day

Institutional guests invited. The honest decision to grow the tool or retire it.

Then

90-day follow-up

Still live? Named owner? Used in the past 30 days? The number that matters most.

07Three worked examples

What a participant actually builds

Public sector

Request tracker

A citizen request passes through six desks and nobody knows where it sits. The participant builds an intake form, status states and an automated flag when a service standard is breached, on top of the existing spreadsheet so nobody has to migrate.

Guardrail Approved infrastructure. Automates no decision a named officer must make.

Health and laboratory

Reading consistency

Prenatal sample readings vary by technician and shift. The participant builds a reading protocol against a reference set, flags outliers between technicians and charts drift by week.

Guardrail Quality assurance only. No patient-identifiable data. The quality manager signs off.

Finance and banking

Memo assembler

Relationship managers rebuild the same credit memo from scratch. The participant builds an assembler drawing from approved sources, drafting to the house template, routing for sign-off with every edit logged.

Guardrail Human sign-off cannot be skipped. No live customer data.

08Readiness & safeguards

Building inside government, health and finance carries obligations

01Policy, law and regulation

No patient, citizen or customer-identifiable data in any student build; synthetic or anonymised only. Data classification is taught in Week 2 and checked at sign-off. Institutional cohorts run inside the partner’s data-protection and procurement rules. Tools record and flag; they do not make regulated decisions.

02Institutional oversight

An accountable owner inside the participant’s organisation is named before Week 1 and signs off before graduation. Institutional cohorts have an internal sponsor. Every tool has a named maintenance owner and a Handover Pack.

03Technology architecture

One documented method, Architecture Before Code, with a specification, data model and permission design before any build. A two-agent workflow: Claude for planning and architecture, Antigravity for execution. Every product in one shared build environment. Standard hosting, standard security review before deploy.

04Leadership and capacity

Two named co-founders who design and teach the programmes. A technical co-instructor for engineering standards and security. Guest tutors from product practice. A domain mentor per track. A cohort manager for roster, progress and reporting.

05Shared practices

A written, reusable method already used by graduates on their own work: the greenfield opener, the 7-part build prompt, ProjectPort, scope files, the Editing Pass, the Evaluation Pass, the Handover Pack. Recaps and handouts ship every session. What we learn is published at principle level.

06Handover, or it is not shipped

A tool without a maintenance owner and a Handover Pack is not shipped; it is abandoned. Both are graduation requirements. Sponsor reporting uses unique landing URLs and cohort-level data, and participant names are shared only with explicit consent.

The lines we hold, whoever is paying

No certificate-first programmes. A live URL stays the graduation requirement.
No enrolments as outcomes. We report deployed tools, users served and 90-day survival.
No identifiable data, no regulated decisions in any student build.
No volume before access. Every builder has a mentor, an accountable owner and a working environment.
No curriculum for money. We take a partner’s problems, tracks and mentors — not direction on the method.

09How impact is made legible

We will make the results verifiable

Every student product moves into a shared KayKav build environment, a GitHub organisation, so a partner can see in one place what was built and by whom, when it shipped and where it is live, and ongoing activity after graduation.

Outcomes are tracked per participant at 30, 90 and 180 days: is the tool still live, how many people use it, and did it lead to a role, a promotion, a customer or a follow-on build.

Proposed indicators for a partner cohort

Seats filled and completed: Enrolled, attended, graduated against the requirements
Tools deployed: Live URLs in the shared build environment at Demo Day
Tools in use at 90 days: Still live, named owner, used within the past 30 days
People served: Colleagues or members of the public whose work passes through a tool
Follow-on outcomes: Roles, promotions, second builds, adoption beyond the unit

Builder intelligence

A partner cohort produces knowledge nobody else is collecting: how domain professionals actually adopt AI to build. Captured with consent, reported at cohort level. Across cohorts this becomes a standing view of what a sector builds first, where it stalls, and what it costs to get people across the line — a public good for anyone funding digital capacity on the continent.

Problem, sector, shapeForm, workflow, dashboard or app
Data reachWhere access was the blocker
Stall pointsWhich phase, skill or guardrail
Tool and model usageThe true cost of inclusion, measured
AdoptionUsers, owner, spread at 30, 90 and 180 days

10Partnership models

Five ways to partner

Fund a block of seats

A partner funds seats in an open cohort for its staff or the communities it supports. Named cohort reporting and a view of every product the funding produced.

Foundations · corporates · associations

Embedded cohort

One institution, one cohort

A closed cohort for one ministry, agency, hospital network or bank, on its approved infrastructure, with its own domain mentor and an internal sponsor who signs off tools.

Governments · health systems · banks

Programme partnership

A capstone for initiatives

Domain Expert Builders runs as the capstone of a training initiative, national programme or fellowship, so participants finish with a deployed tool.

Public programmes · fellowships · NGOs

Hybrid hub

A partner-hosted space

A hub or training room with devices, connectivity and power, where a cohort attends live sessions together and builds on site. Removes the access barriers that exclude people before pricing does.

Hubs · state agencies · universities

Tool partnership

Access for outcomes

AI and developer tool providers supply educational access and credits in exchange for usage and outcome data from people who ship.

Technology companies

The split

Partner provides

Funding for seats and tool usage
An internal sponsor
Participation at Demo Day

KayKav provides

The full six-week programme
The method, toolchain and build environment
Guardrail design for the sector
Cohort management and named reporting
30, 90 and 180-day tracking

11What it costs to include people

Tuition is not the real cost

Two cohorts have taught us that the largest running cost is AI model and tool usage per participant. Behind that sit laptops, mobile data and stable power, which exclude capable people before pricing does.

A seat that covers only tuition leaves a civil servant or laboratory scientist unable to complete. A fully inclusive seat has four lines.

Tool cost has a known lever: education-tier pricing negotiated at network level, so a sponsor’s tool-credit pool goes further each cohort. Seat pricing with volume tiers on request.

01TuitionThe programme, mentoring, reporting and the graduation standard

02Tool creditsAI model and tool usage for six weeks of directed building

03Data stipendWhere connectivity is the blocker to attending and building

04Device accessWhere no laptop is available to the participant