Lessons and tools from Kabakoo’s WhatsApp learning system
Kabakoo Academies · September 2026
An AI response is a beginning. What can someone do with it?
Follow Kabakoo’s WhatsApp Learning System from an invitation to feedback, peer support, and work beyond the course. The aim is economic mobility. Use these lessons to build and test your own service.
Follow the work beyond the screen.
A daily task on WhatsApp. Feedback to work with. People to turn to. Kabakoo connects these parts around learning that can travel into someone’s working life.
6,671learners engaging through WhatsApp
January–June 2026
826learners enrolled in the first Tremplin cohort
Launched March 2026
Tremplin begins with a 30-day WhatsApp challenge: one short task each day, with the AI Mentor answering questions and reviewing work. The wider learning experience combines WhatsApp with app-based courses and portfolios, peer exchanges, and in-person sessions.
Facilitators and mentors support the work in Kabakoo’s learning community. Peer Pods are being tested as a route into sustained collaboration. The 2026 service · Follow the learning sequence ↗
MAÏMOUNA · STORYTELLING
Maïmouna describes AI feedback that showed her what to add or change and helped her submit videos. Her wider Kabakoo journey includes storytelling and digital marketing for her hair care brand, Aÿroün.
Follow her learning journey ↗
SOUMAÏLA · MAKING AND REPAIR
Soumaïla brought phone-repair experience into Kabakoo’s hybrid programs. New machine-hacking skills helped him repair the electronics of a pottery kiln that another business needed.
Follow the work into the workshop ↗Photographs: Kabakoo Academies.
In Kabakoo’s broader hybrid programs, the 32-month follow-up found a stronger income profile among participants: 11% reported earning above 250,000 FCFA per month, compared with 7% in the comparison group. The share in the lowest bracket was also smaller, and participants reported more investment in their working lives.
These findings concern the earlier hybrid programs. The 2026 WhatsApp cohort has its own learning and follow-up journey. Read the study context · What we will explore next ↗
The 2024 Mali youth trust survey put WhatsApp first among institutional contact channels: 21.5% of 1,079 respondents. It prompted Kabakoo’s turn toward WhatsApp. Explore the survey ↗
Jamie Walsh’s onboarding analysis directs attention to the invitation, sender, timing, and first action. Those are design decisions a team can test before adding more content. Examine first use ↗
Elia Gandolfi’s analysis associates in-person participation with more WhatsApp activity. That makes earlier human contact a useful next experiment. Investigate continuation ↗
This playbook draws on more than six years of building and learning with AI at Kabakoo. In April 2020, members of our community were already working on image labeling and building AI solutions together. That December, our UNESCO workshop with Abeba Birhane and André Uhl examined how AI carries the values and assumptions of those who build it. We launched the first text version of our AI Mentor in March 2023 and introduced Bamanankan voice interaction in March 2024. Those experiences sit behind the WhatsApp learning system described here.
The World Economic Forum named Kabakoo among 16 “Schools of the Future” in 2020. In 2024, it featured Kabakoo among nine case studies in Shaping the Future of Learning: The Role of AI in Education 4.0, highlighting our use of AI mentoring to support young people’s learning and livelihoods.
This playbook shares the knowledge accumulated through that work, including the decisions we revised and the questions we are still investigating.
Read in order, or choose the decision your team needs to make next.
Ten connected decisions, from purpose to continued learning.
See the knowledge already present. Find one obstacle your application could address.
Frame the problem ↗ 02 / CHANGEFollow the actions and relationships between a useful response and a better life.
Explore the four evaluation levels ↗ 03 / RESEARCHTest on a real device, observe the pauses, and decide what the next pilot needs.
Prepare a research round ↗ 04 / LEARNINGMake the attempt, feedback, peer contribution, and revision work together.
Walk through the experience ↗ 05 / ARCHITECTUREExplore n8n, Django, HAKILI, storage, delivery, and recovery when a dependency fails.
Open the engineering walkthrough ↗ 06 / AI MENTORReview responses, trace multilingual voice, and examine what learners do next.
Test the mentor’s job ↗ 07 / FIRST USEUse the onboarding analysis to investigate the gap before the first response.
Explore the onboarding evidence ↗ 08 / CONTINUATIONFind the obstacle behind an interruption and preserve a usable route back.
Investigate a stalled task ↗ 09 / RELATIONSHIPSConnect matching, agreement, and facilitation to help people can use.
Try the matching priorities ↗ 10 / EXPERIMENTSInspect assignment, delivery, outcomes, and the decision a result supports.
Explore the denominators ↗
Soumaïla was studying civil engineering, repairing phones, and helping his mother sell fish. At Kabakoo he developed machine-hacking and 3D-printing skills, then put them to work repairing the electronics of a ceramic kiln. New learning extended skills he was already using to earn a living. Read Soumaïla’s story.
The Agency Fund’s four-level evaluation framework gives us a shared language: Model, Product, User, Impact. Kabakoo’s essay, The AI worked. Did it work for the learner?, asks what those levels reveal—and whom they can miss. A good model response, continued participation, changed capability, and improved livelihoods each need their own evidence.
Explore the framework through one task · Read the sources · About the web edition
How does learning contribute to economic mobility? Who governs the data, operates the service, and pays for delivery? What should change when the system moves somewhere new?
Chapters 11–15 will examine outcomes and agency, governance, team operations, costs, and adaptation. Explore the next chapters and request early access.