Africa’s AI Class of 2025, and the Gap It Left Behind

By Darlyn Okojie

A year of launches from Lagos to Addis Ababa has put AI into banking, hiring, software and commerce. What’s still missing is a way to know whether any of it will stick.

By most measures, this has been the year African tech got back on its feet. Startups on the continent have raised more than $3 billion, the highest total since the end of the zero-interest era in 2022, and investors have returned to companies with real revenue and a clear route to profit. Much of the new energy has gone into artificial intelligence. At least sixteen new AI startups have been founded on the continent this year, adding to a base that has nearly doubled since 2022.

The launches covered most of the economy. In Ethiopia, Gebeya released Dala, an AI app builder designed for African users and developers. In Nigeria, Sulaiman Adewale launched Xara, a WhatsApp banking assistant trained to understand Nigerian English and Pidgin, with Hausa and Yoruba to follow.

Tunisia’s Karim Jouini and Jihed Othmani, the team behind Expensya, returned with Thunders, a platform that uses AI to take the tedium out of software testing. ALX launched Chidi with Anthropic and the Rwandan government, and in Ghana, JobPilot AI set out to give job seekers a single AI companion for listings, CVs and interview preparation.

Commerce and marketing moved too. Yelen began building online selling infrastructure for small businesses across Francophone Africa that had been shut out of e-commerce. In Lagos, Faith Aminu launched Kave Africa, a platform connecting brands with nano and micro creators, which drew 3,500 creators to its first festival this month. At the enterprise end, AfricAI, a joint venture with Nigeria as its flagship market, set out to build AI for healthcare, digital identity, document automation and public administration.

Across established companies, the same pattern has played out. African businesses have spent the year putting generative AI to work in customer support, content, compliance and internal operations.

The layer nobody built
Look closely at that list and a gap appears. Nearly every launch this year is a tool: something an organisation buys, installs or rolls out. None of them answers the question every board eventually asks, which is whether the people expected to use the new system actually will.

That isn’t a small question. In August, an MIT study found that 95 per cent of corporate generative AI pilots stall before reaching scale, and its authors put the blame less on the technology than on how organisations bring it in. Fragmented workflows, weak change management and staff quietly using unsanctioned tools all play a part. As African companies move from experimenting with AI to depending on it, the same risk follows them.

One African-founded company has made that gap its entire business. Humaine Solutions, founded by Darlyn Okojie and based in Manchester, builds HTIE, which it calls an adoption intelligence layer. Its starting premise is simple. Every serious transformation programme tracks budget, governance and technical delivery in real time, but adoption usually goes unmeasured until go-live, by which point problems are expensive to fix.

HTIE gives each stakeholder group and site an Adoption Risk Score before rollout begins, drawing on organisational network mapping, change-fatigue history and readiness diagnostics. During rollout it tracks that score through short pulse surveys, usage data and feedback from internal champions, the way a programme office tracks budget. When a score drops, it identifies which of four drivers is failing (autonomy, complexity, reward or governance) and points to the matching fix. If the problem is complexity, the answer might be a simpler workflow rather than another round of training.

The model is grounded in an analysis of 25 major transformation failures, including GE Predix, IBM Watson Health and the NHS National Programme for IT, each traced to the same missing layer. HTIE is now being validated through a three-phase pilot that runs alongside an organisation’s existing change tracking.
Why it matters for Africa.

The case for adoption intelligence may be stronger in African markets than anywhere else. Banks, telecoms and government agencies across the continent are layering AI onto workforces that span very different levels of digital fluency, often across several countries and languages at once. A chatbot that works in Lagos can fail in Kano for reasons no technical dashboard will show. A national digital identity system can be technically sound and still go unused.

This year has proved that Africa can build AI products for its own context. The question it leaves behind is whether the organisations buying them can make them land.

What 2026 will bring
If this has been the year of the launch, 2026 is likely to be the year of the reckoning. Several signals already point that way.

First, the money will keep coming, but on harder terms. The investors who returned this year are rewarding revenue and discipline over growth at any cost, and mergers and acquisitions across African tech have reached a record. Expect 2026 to bring fewer but larger funding rounds, and more consolidation as weaker startups sell to stronger ones rather than raise again.

Second, corporate AI budgets will rise faster than the ability to measure them. Banks, telecoms and retailers that piloted generative AI this year will move to larger deployments, including AI agents that carry out tasks on their own. Most will not yet have a reliable way to track whether that spending is paying back, and boards will start asking for proof of return in a way they have not so far.

Third, AI will begin to show up in headcount decisions. As customer support and back-office work is automated, African companies will start citing AI openly when they restructure. Each of those decisions will rest on an assumption that the new system has been adopted well enough to replace what it displaced.

Fourth, the gap between people and organisations will widen. Africa’s young, digitally fluent workforce is picking up AI tools faster than most companies are redesigning around them. In markets such as Nigeria, the risk in 2026 is not a shortage of AI-ready talent but organisations that are not set up to use it.

Taken together, these signals point to one conclusion. The advantage in 2026 will belong to organisations that measure adoption, not just technology. Those that can see before go-live where a new system will stall, and why, will turn this year’s launches into results. For the tools built in 2025, including those like HTIE that were designed for exactly this problem, 2026 will be the first real test.