How Tathmini Africa is using AI to help Kenyan families navigate the choices created by CBE

For many Kenyan families, choosing a senior school pathway is becoming a complicated exercise in information, interpretation and timing. Learners moving through the Competency-Based Education (CBE) system must make decisions about pathways and subjects at an age when many parents are still trying to understand how the new curriculum differs from the 8-4-4 system they experienced themselves.

Tathmini Africa is betting that AI can help make those decisions easier, provided the technology is built around the realities of the Kenyan education system.

The startup, founded by Isaac Muchembe Kihumba, combines learner assessments, parent and teacher input, academic performance and career expectations with information about CBE pathways, subject combinations and senior schools. Its approach reflects a wider shift underway in Kenya’s technology sector, where local developers are increasingly trying to build AI and EdTech products around local data, connectivity constraints and practical needs rather than simply adapting tools created for other markets.

For Kihumba, the idea started with a personal experience.

He spent two years studying Engineering at Kenyatta University after his KCSE results qualified him for the course, before switching to Economics and Finance. The decisions, he says, were made without much tangible guidance.

Years later, he watched his niece choose her Senior School subjects.

At 13, she was confronting a problem Kihumba had experienced more than a decade earlier, although under a different curriculum.

That became the starting point for Tathmini Africa.

CBE has made career guidance an earlier decision

The challenge facing families is partly a consequence of how early important decisions now arrive.

According to Kihumba, learners around Grade 9 are typically 14 or 15 and must begin making choices around three broad pathways: Science, Technology, Engineering and Mathematics (STEM), Arts and Sports Science, and Social Sciences.

Those pathways then branch into different subject combinations, while individual schools offer different combinations, locations and boarding arrangements.

Kihumba says Tathmini’s database currently covers more than 400 Ministry-approved subject combinations across more than 10,000 senior schools. Those figures should be read as the company’s description of the education data it has mapped and independently checked against the latest Ministry records before publication.

For families, the underlying problem is easier to understand than the numbers suggest. A learner may have a strong interest in a particular set of subjects, only for the family to discover that the combination is unavailable at the schools they can realistically consider.

Parents are also making these decisions with varying levels of familiarity with CBE.

Many went through 8-4-4 and have little personal experience with the pathway structure now facing their children. Others may fall back on familiar career choices that worked for relatives under a completely different education system.

That can leave learners making decisions based on incomplete information, family expectations or simple guesswork.

Tathmini’s proposition is to bring those pieces together and turn them into a narrower set of options that a family can investigate.

Tathmini starts with three views of the learner

The platform does not base its assessment on the learner alone.

Tathmini uses what Kihumba describes as a tri-perspective approach in which the learner, a parent and a teacher independently complete a 40-question assessment.

The assessment is built around RIASEC, or the Holland Codes, a career-interest framework that groups interests into six broad categories: Realistic, Investigative, Artistic, Social, Enterprising and Conventional.

A learner can therefore emerge with a combination of interests rather than being assigned a single career identity. Someone might, for example, receive an “SI” profile, reflecting Social and Investigative interests.

Tathmini then combines that profile with academic performance and career expectations before mapping the information against CBE pathways, subject combinations and senior schools.

That distinction is important.

The platform is not simply trying to answer the question, “What career should this child pursue?” It is attempting to provide several possible routes and show how a learner’s interests and circumstances relate to them.

The involvement of parents and teachers is also intended to make disagreements visible.

A learner may see themselves differently from a parent. A teacher may have observed abilities that neither has considered. Bringing the three perspectives together creates another conversation around the recommendation rather than allowing an algorithm to make the decision on its own.

The local data problem behind AI in education

This is where Tathmini’s proposition connects with a broader issue emerging across Kenya’s technology sector.

The country has rapidly expanded access to digital services, AI tools and online learning. But access to a powerful general-purpose model does not automatically give a user an accurate answer to a highly local question.

Ask a general AI chatbot which senior school in a particular part of Kenya offers a specific CBE subject combination, for example, and the model may not have reliable, current information. It can produce a plausible response without having a dependable underlying database to verify the answer.

For education, that distinction matters.

Tathmini says it built its school and subject-combination database from Ministry of Education records, mapping senior schools, approved combinations and the relationship between schools and combinations down to the sub-county level. Its career mapping is then connected to subjects under Kenya’s CBE system and the requirements of Kenyan institutions.

The company says its AI counsellor uses that local information when answering questions.

That approach mirrors a broader direction in Kenya’s AI market. The technology conversation is increasingly moving away from simply asking whether a model is intelligent enough and toward whether the data, systems and context surrounding the model are reliable enough to produce useful results.

The same issue has emerged in enterprise and government technology, where fragmented systems and disconnected datasets can limit what AI agents can actually do.

Education has its own version of that problem.

A model can know a great deal about medicine, engineering or architecture globally. It still needs accurate Kenyan information to explain which secondary-school subjects lead toward those fields, which institutions have particular requirements and which schools currently offer the relevant combinations.

Building AI around Kenya’s constraints

Tathmini’s local approach extends beyond its database.

Kihumba says the platform is designed to work through a browser on an ordinary smartphone rather than requiring an app download. The company has also prioritised low-data usage and M-PESA payments, with reports starting from KSh50, according to the founder.

Those choices fit a broader pattern visible in Kenya’s EdTech market.

Local education technology companies are increasingly designing around the realities of inconsistent connectivity, data costs and uneven access to devices. Other Kenyan EdTech startups have pursued offline-first learning, while government programmes are expanding digital devices, smart classrooms, connectivity and digital skills across schools.

But infrastructure alone does not remove the access problem.

A school may have digital equipment while a learner’s family still relies on mobile data. A student may have access to an AI tool but lack the knowledge to judge whether its answer is reliable. A parent may have a smartphone but remain unfamiliar with CBE pathways.

For Kihumba, building for Kenya therefore means accounting for the way technology is actually used in Kenyan households.

“A child’s future here is a family conversation,” he says.

That is why parents and teachers are part of Tathmini’s assessment rather than being treated as observers of an AI-generated result.

AI can widen access to career guidance, but it can also widen inequality

Kihumba sees a larger opportunity in making career guidance available to families that might otherwise struggle to access it.

Historically, families with greater financial resources have been able to pay for counsellors, aptitude assessments, school visits and wider exposure to careers. Learners in under-resourced schools may receive considerably less guidance.

AI could reduce some of that gap by making basic personalised guidance available at a lower cost.

But the technology can just as easily deepen inequality if AI products assume continuous broadband, expensive subscriptions or access to devices and support that many families do not have.

This is one reason low-data access and mobile-money payments are central to Tathmini’s proposition.

It is also why Kihumba rejects the idea that AI should replace teachers, parents or career counsellors.

“AI can be the great equaliser or the great divider,” he says. “It depends entirely on how it’s designed.”

That argument is increasingly relevant as Kenya expands AI training and digital-skills programmes. Recent initiatives have taken AI education into public libraries and underserved communities, while universities, technology companies and workforce programmes are trying to prepare young people and professionals for an economy in which AI skills are becoming more important.

The question is shifting from whether Kenyans can access AI to whether they can use it meaningfully.

For younger learners, that includes understanding what their own interests and abilities might mean in a labour market that is changing rapidly.

A recommendation should open possibilities, not close them

There is an obvious risk in using AI to advise children about their futures.

An algorithmic recommendation can quickly become an authority figure, particularly when parents are uncertain about the education system themselves.

Tathmini’s response is to present recommendations as possibilities rather than verdicts.

Its reports provide several potential pathways and the reasoning behind them. The disagreement between learner, parent and teacher can also appear in the report, creating an opportunity for further discussion.

The company’s AI counsellor can direct users toward human counsellors and teachers, while reports include disclaimers and links to official government portals so families can verify information.

Kihumba describes the product as “a compass, not a GPS.”

That framing is important because a learner’s interests are not static. A 14-year-old who is strongly interested in one subject today may develop a different interest after encountering a new teacher, activity, technology or career.

Tathmini therefore encourages families to revisit assessments rather than treating one result as a permanent label.

The same principle applies to academic performance.

Kihumba says the platform is designed to avoid automatically steering a learner away from ambitious pathways simply because their grades are average. Interests are given significant weight, while the learner, parent and teacher perspectives provide multiple inputs instead of allowing one person’s assumptions to determine the outcome.

No assessment system is free from bias. The more consequential question is whether the product makes its assumptions visible and gives families room to challenge them.

Kenya’s local EdTech sector still faces a scaling problem

The technology may be the easier part of the equation.

Getting local EdTech products into the hands of thousands of learners is considerably harder.

Kihumba identifies several barriers: cautious parents and schools, the cost of reaching schools across all 47 counties, limited household and school budgets, demand that peaks around selection periods, and difficulty raising capital for products that may require significant time to build trust and scale.

Public-sector procurement is another challenge.

Kenya has invested heavily in digital education infrastructure, but local startups can still struggle to find a clear route from pilot programmes to large-scale adoption.

Kihumba argues that more structured pilots involving the Ministry of Education and county education offices could give locally developed products a route to demonstrate their usefulness.

There is also a data question.

As more government education information becomes digitised and published, local developers have a better foundation on which to build. But the usefulness of those datasets depends on their completeness, consistency, accessibility and frequency of updates.

For a product making recommendations about schools and subject combinations, outdated information can quickly undermine user confidence.

That makes data maintenance just as important as the AI layer.

The real test is whether local context produces better decisions

Kenya’s technology ecosystem is entering a period in which AI is becoming increasingly embedded in education, workforce development, enterprise software and public services.

At the same time, the country is still dealing with the older problems of connectivity, affordability, fragmented information and unequal access.

Tathmini sits at the intersection of those two realities.

Its proposition is relatively straightforward: use AI to help families make sense of a complicated education system, but ground that AI in Kenyan data and keep teachers, parents and learners involved in the decision.

Whether that model works at scale will depend on factors beyond the technology itself. The accuracy of school and subject data, the quality of the assessments, the reliability of recommendations, user trust and the company’s ability to reach families outside the most connected parts of the country will ultimately determine its value.

For Kihumba, however, the motivation remains personal.

He studied Engineering because his grades made him eligible. He later moved into Economics and Finance before eventually finding his interest in technology.

His concern is that another generation could make similarly consequential choices without understanding where their interests might lead.

“Every Kenyan child carries a spark,” he says.

The ambition behind Tathmini is to help families identify it earlier, with better information and more room for the child to shape the answer.

Real ESG impact doesn’t happen in panels alone, it happens in the rooms where financiers, operators, and policymakers actually align. Our GreenShift Forum 2026 cuts the noise, bringing together the people rewiring Africa’s sustainability and energy frameworks for one focused day in Nairobi. Secure your seat.

Go to TECHTRENDSKE.co.ke for more tech and business news from the African continent and across the world.

Follow us on WhatsApp, Telegram, Twitter, and Facebook, or subscribe to our weekly newsletter to ensure you don’t miss out on any future updates. Send tips to info@techtrendsmedia.co.ke