01 The problem: learning that doesn't become a career
Climate leadership training in Africa faces a gap that no amount of content fixes. A learner finishes a module, feels informed — and then hits the questions that actually decide their future: how do I turn this into work? What should I learn next? Who is hiring, and how do I present myself to them? On most learning platforms those questions go unanswered, because answering them at scale, in the learner's own language, at the hour the learner asks them, is an operations problem no teaching team can staff.
ACLA built Tapi Learn to close that gap: a free, mobile-first platform for climate-leadership learning that treats the learner's career — not just their coursework — as the product. My mandate was the AI layer that makes that possible: guidance that is always on, speaks three languages, and knows each learner's situation well enough to be useful.
02 The constraints that shaped the build
- Three languages, real users. English, Français, and Kiswahili aren't decoration here — they're how thousands of learners across 38 countries actually write. The NLP pipeline had to treat all three as first-class, including Kiswahili, which most global AI tooling treats as an afterthought.
- Coaching can't keep office hours. Learners across 38 time zones need answers at 2 a.m. before an interview. A human-coaching model doesn't scale to that; a good AI layer does.
- One coach, thousands of paths. Personalization across highly diverse learners — different countries, experience levels, goals — without every recommendation being hand-built.
- Free platform, real costs. The product is free to learners, which means every AI interaction has to earn its keep. Response paths were designed to be useful at low cost, not to burn tokens for sport.
- Mobile-first, low-bandwidth. Many learners are on mid-range Android phones with intermittent data — the same delivery conditions I design for across all my Africa-facing builds.
03 What I built
The 24/7 AI Career Coach
CV reviews, interview preparation, and career advice on demand — the guidance layer that turns "I finished the course" into "I'm ready to apply." It works in the learner's language, at the moment they need it, and it never closes.
The EN·FR·SW NLP pipeline
A multilingual pipeline that understands and serves learners in English, Français, and Kiswahili — the language layer that decides whether a learner feels at home on the platform or talks past it.
Personalized learning recommendations
Recommendations shaped by where a learner is and where they're heading, so "what should I learn next?" has a real answer instead of a generic syllabus order.
The opportunities board
The bridge from learning to livelihood: real opportunities surfaced to learners, so the platform's last screen is a door — not a dead end.
Most learning platforms bolt a chatbot onto a course catalog and call it AI. Tapi Learn's AI layer was designed the other way round: start from the learner's career problem — guidance, language, next steps, opportunities — and work backwards to the features. That's why the coach feels like the product instead of a support widget.
04 Results: reach, resonance, and a live platform
| Dimension | Where it stands (Oct 2026) | Why it matters |
|---|---|---|
| Reach | 11,000+ learners across 38 countries (client-reported) | Multilingual, mobile-first AI isn't a nice-to-have — it's what got the product across borders. |
| Satisfaction | 4.8/5 average rating (client-reported) | Learners rate the whole experience — the AI layer carries a real share of that. |
| Always-on guidance | 24/7 career coaching in 3 languages | The platform answers the career question at the moment it's asked — not in office hours. |
| Live product | Public at acla.io/tapi-learn — a free platform with a real operating model | The system runs in production, serving thousands of real sessions. |
Tapi Learn needed guidance that never closes and speaks our learners' languages. Victor delivered a 24/7 Career Coach in English, Français and Kiswahili — and the product visibly moved: 11,000+ learners across 38 countries and a 4.8/5 rating. Learners treat the coach as part of the product, not a support widget.
What this means for your business
- If you run a learning, membership, or community product — the retention math changes when every user gets guidance, not just content. Tapi Learn shows what an AI guidance layer does to reach and satisfaction in production.
- If your users are multilingual — treating their language as first-class isn't localization cost, it's market expansion. Three languages took one product into 38 countries.
- If you've priced out human coaching/support — a well-scoped AI layer delivers the always-on version at a fraction of the staffing cost, and this one has the production evidence to prove it.
Want AI that carries your product experience?
Book a 30-minute AI systems assessment. Bring your product and your users — you'll leave with an honest read on where an AI layer would earn its keep, and where it wouldn't.