From Prototype to Investable Company: What African AI Founders Must Prove

By Sir Roger Jantio, Senior Managing Director & CEO, Sterling Merchant Finance Ltd, Washington, DC

This is the second article in a three-part series on Africa’s AI investability challenge. The first article argued that Africa’s AI debate is no longer about awareness, but investability. 

This turns to founders. It is rooted in the author’s AI investment experience with a broad range of founders across multiple markets, and in his investment approach to identify companies capable of becoming standout companies before that becomes obvious to everyone else.

The central point of this article is simple: a strong demo, a promising pilot or a technical breakthrough can open the door. But serious capital will look for more: a painful problem, a clear customer, evidence of willingness to pay, defensible data, credible economics, a capable team and a path from first use case to repeatable growth.

Investors do not finance enthusiasm. They finance evidence. That does not mean African founders should become timid. Quite the opposite. Africa needs entrepreneurs willing to solve hard problems in agriculture, healthcare, education, finance, logistics, energy, insurance, trade, public services and small-business productivity. It needs AI founders who want to create companies that can outgrow their first market and become category leaders.

But ambition becomes investable only when it is translated into proof. For African AI founders, ten tests matter most.

The first test is problem clarity.

Too many AI projects begin with the technology. A founder discovers a model, builds an interface, creates an application, and then searches for a customer. That path can produce interesting experiments, but it rarely produces durable companies.

The better path begins with a painful problem. What is broken? Who suffers from it? How much does it cost? Who is already paying for a partial solution? Why is the existing solution inadequate? Why is AI necessary? Why now?

In African markets, this discipline matters even more because many problems are real but not all are immediately monetizable. A farmer may need crop advisory services, but may not be the direct payer. A patient may need diagnostic support, but the clinic, insurer, donor program or government may control the budget. A school may benefit from an AI tutor, but the parent, school operator, ministry or development program may be the paying customer. The founder must know the difference between the user, the customer and the payer.

The second test is that AI must be core to value creation.

Not every company using AI is an AI company. Some are ordinary software businesses with an AI feature. Some are service companies using AI internally. Some are wrappers around third-party models with limited defensibility. Some are marketing stories designed to benefit from investor excitement.

There is nothing wrong with using external models, APIs or open-source tools. Many strong companies will be built that way. But founders must be honest about where value is created. Is the company valuable because of the model itself, the proprietary data, the workflow integration, the distribution channel, the regulatory position, the user experience, the sector expertise, or the ability to deliver measurable outcomes?

Founders should be able to show what improves because their product exists. Does it improve yields? Reduce diagnostic time? Expand access to credit? Lower fraud? Reduce logistics delays? Improve learning outcomes? Increase energy efficiency? Reduce claims leakage? Improve public-service delivery? Save money for customers? Increase revenue for small businesses?

Broad claims are not enough. “We improve healthcare” is weaker than “we reduce diagnostic turnaround time by 40 percent in district clinics.” “We support farmers” is weaker than “we improve crop disease identification and reduce treatment delay by five days.” Customers pay for measurable value. Capital follows clarity.

If the underlying model can be replaced tomorrow and the customer would barely notice, the company may not be defensible. Defensibility in African AI will often come from local data, domain knowledge, trusted relationships, regulatory understanding, distribution networks, language capabilities, low-bandwidth deployment, integration into existing workflows and the ability to operate in fragmented markets.

The third test is customer evidence.

Founders should be careful with the word “pilot.” A pilot can mean many things. Some pilots are paid. Some are unpaid. Some are procurement tests. Some are donor-funded demonstrations. Some are polite experiments that will never become contracts. Some are serious first steps toward enterprise adoption.

Investors will ask: who is paying? How much? For how long? What must happen for the pilot to become a contract? Is there a budget line? Who signs? What is the procurement cycle? What are the implementation risks? Has the customer used the product repeatedly? Is the customer willing to expand usage?

A founder who says, “we have five pilots,” has not said enough. A stronger founder says: “We have two paid pilots, one converted into a twelve-month contract, three enterprise customers in procurement, a six-month sales cycle, and early evidence that customers save 25 percent in processing time.” That is a different conversation.

The fourth test is revenue quality.

For venture-backed companies, revenue is not just revenue. Investors look at the quality, predictability and scalability of revenue.

Is there monthly recurring revenue? Is there annual recurring revenue? What is the annual contract value? Is the revenue transactional, subscription-based, usage-based or project-based? Is revenue concentrated in one customer? Are customers renewing? Are they expanding usage? Is the company dependent on grants or one-off implementation fees?

An AI startup selling to enterprises or institutions must understand the language of recurring revenue, retention, pipeline, conversion, churn and expansion.

A company with modest early revenue but strong retention, clear customer pain, improving margins and a growing qualified pipeline may be more investable than a company with a large one-time contract and no repeatable sales motion.

The fifth test is unit economics.

This is especially important in AI. An AI product can be impressive and still be economically weak. Compute costs, API costs, cloud hosting, data processing, model training, inference, human review, customer support and implementation can all affect margins.

Founders must know their gross margin after compute and delivery costs. They must understand whether each additional customer improves the economics or makes the business more expensive to operate. They must know whether they can reduce costs through model optimization, edge deployment, better architecture, local hosting, batching, caching or workflow redesign.

If a company spends too much money to deliver each unit of AI value, scale can make the problem worse, not better. This is where many AI founders globally struggle, not only in Africa. But African founders must be especially disciplined because capital is scarcer, infrastructure can be less reliable, and customers may be more price-sensitive.

The sixth test is infrastructure realism.

This is the point many outside investors underestimate and many African founders live every day. An African AI founder may be judged against global standards while operating with frictions that founders in more developed ecosystems rarely have to solve: unreliable electricity, unstable connectivity, high cloud costs, fragmented data, paper records, informal workflows and customers who may lack the digital infrastructure required to adopt the product easily. That is not a talent problem. It is an operating-environment problem.

In many African markets, the founder is often building two things at once: the product company and part of the environment needed for the product to work. This creates what may be called the African execution premium: the hidden cost of building where infrastructure cannot always be assumed. It does not excuse weak execution, but it changes how execution should be understood.

A serious founder cannot simply complain about the constraint. A serious investor cannot pretend the constraint does not exist. The investability question is whether the founder has designed the business to survive and scale in the real market. Can the product work in low-bandwidth environments? Can parts of the system function offline? Can it integrate into messy workflows? Can it produce value before the whole environment is perfect?

African founders who answer these questions well are not weaker than founders elsewhere. They may be more resilient. If they can build products that work under constraint, they may eventually build companies with deep defensibility.

The seventh test is data rights.

Data is not an afterthought. It is often the heart of the company. What data is required for the product to work? Where does the data come from? Who owns it? Who has the right to use it? Is consent required? Are there privacy obligations? Can the data be transferred across borders? Can it be used to train or fine-tune models? Is the data exclusive, proprietary or easily replicable? What happens if a partner withdraws access?

In sectors such as health, finance, insurance, education, agriculture and public services, data rights can determine whether a company is investable or fragile. Health is especially instructive. There is a difference between data that has merely been collected and data that can be trusted enough to build on. Bad data is not just inefficient. It can be dangerous.

African founders should not treat data governance as a legal burden to be addressed later. It is part of company value. A founder with clear data rights, credible privacy practices and trusted institutional relationships is building an asset. A founder relying on unclear access to sensitive data is building risk.

The eighth test is scale.

For many African AI companies, the first serious test is more basic: can the product move beyond one pilot, one department, one institution, one city, one customer group, or one narrow use case?

A good African AI company should be built with local understanding, but not trapped in a one-off deployment. Many promising startups fail not because the first use case is weak, but because the solution cannot be repeated economically. It works with one customer because the founder is personally involved. It works in one location because the team manually adjusts everything. It works in one pilot because the implementation is subsidized. That is not yet scale.

Founders must therefore understand what can be repeated. Can the product serve a second customer without being rebuilt? Can it move from one branch to ten branches? From one clinic to a network? From one school to a school system? From one crop to adjacent crops? From one claims workflow to another? From one customer segment to a larger segment with similar needs?

The answer will differ by company. Some AI businesses will scale through enterprise customers. Others will scale through sector partners, distribution channels, institutional buyers, or repeated deployments within a large customer base. The point is not that every company must become pan-African immediately. The point is that the founder must show a credible path from first proof to repeatable growth, one where the problem, data, customer, talent, regulation and distribution path create the strongest early wedge. 

The ninth test is team quality.

Capital follows teams. A strong African AI company needs more than one brilliant technical founder. It needs technical capability, sector understanding, commercial execution, product discipline, governance and the ability to sell. In regulated sectors, it needs credibility. In enterprise markets, it needs patience and implementation capacity. In consumer markets, it needs distribution. In infrastructure-heavy models, it needs operational competence. Investors ask a simple question: can this team execute the next stage?

At pre-seed, the team may be incomplete. That is acceptable. But the founder must know what is missing and how to fill the gap. At seed, the company should begin showing sharper evidence of product, customer discovery, early revenue or strong validation. By Series A, the market expects more: repeatable sales, stronger metrics, clear growth, improving unit economics and a credible path to scale.

Founders should not confuse fundraising stages. Pre-seed capital is often about team, insight, problem depth and early validation. Seed capital is about building proof. Series A is about showing repeatability. Growth capital is about scale.

The tenth test is choosing the right capital.

Not every AI company is venture-backable. That is not an insult. It is a financing reality. Some companies may be excellent businesses but not suitable for venture capital because they grow more slowly, require heavy implementation, depend on project revenue or operate in markets where exits are uncertain. Others may be better suited for strategic investors, development finance, project finance, grants, revenue-based financing, corporate partnerships or blended capital.

Founders must understand what kind of capital matches their company. Venture capital looks for large markets, rapid growth, strong margins, defensibility and the possibility of major outcomes. Development finance institutions look for commercial viability, additionality, governance, development impact and the ability to mobilize private capital. Strategic investors look for integration, market access, technology advantage or sector positioning. Governments may care about public-service outcomes, national capacity and local value creation. A founder who understands capital fit raises better and wastes less time.

The purpose of these tests is not to discourage founders. It is to prepare them.

Africa does not need timid entrepreneurs. It needs founders who can sit across from investors, customers, regulators and institutions and explain why their company matters, why it can scale, why it is defensible and why now is the time to build. Founders seeking serious capital should be prepared to present evidence, not merely ambition: problem clarity, customer traction, revenue quality, data rights, unit economics, infrastructure realism, team capacity and a credible path from first proof to repeatable growth. That is the standard. It is also the opportunity.

Some of Africa’s future AI category leaders, and perhaps future unicorns, may not look obvious today. They may be working on crop intelligence, clinical workflows, credit infrastructure, local-language AI, logistics optimization, energy management, insurance automation, trade finance, compliance, education, SME productivity or public-sector efficiency. They may begin in small markets. They may start with unglamorous workflows. They may not yet have perfect pitch decks.

But if they solve painful problems, control or lawfully access valuable data, produce measurable outcomes, build strong teams and learn to speak the language of capital, they can become investable.

The African AI founder’s task is therefore not to sound fashionable. It is to become financeable. That is the journey from prototype to company. And that is where Africa’s AI future will begin to separate from the noise.

Roger B. Jantio is an AI investor and strategic advisor focused on artificial intelligence, development finance, emerging markets, and strategic capital. He is the founder and CEO of Sterling Merchant Finance Ltd, a Washington-based merchant bank active across Africa for more than three decades, and of its affiliated investment funds.