The Silicon Savanna is at a crossroads. On one hand, we have the battle-hardened veterans who remember the raw, manual grit it took to build software just a few years ago. On the other, we have a new generation of devs armed with AI tools that can spin up infrastructure with a single prompt.
Watch Now!
In the latest episode of The Tech Room, host Eddie Kareera sat down with Vincent Omondi, Co-Founder and CTO of PayHero Kenya and a seasoned data engineer, to bridge this generational gap.
From gruesome infrastructure “war stories” to the imminent danger of generic AI wrappers, Vincent provided a masterclass on how to build technical utility that actually lasts in Kenya.
Here are the four key insights from their conversation.
1. The Value of Doing Things the Hard Way
Today, spinning up a virtual machine is often just a prompt away. But back in 2019, following a surge in users after a KTN interview, Vincent and his team at Pay Hero spent three sleepless days manually setting up Ubuntu servers, firewall rules, and replication environments.
While modernized, streamlined processes are necessary, Vincent argues that the modern lack of manual experience is a double-edged sword.
“It’s very gruesome, but it gives you the mental model in terms of how the structure works,” Vincent explained. “With modern tools that exist, especially with AI, it has limited most of those particular stuff… debugging becomes a thing that is very, very painful because I don’t know what has broken.”
To build trust in fintech, you have to know your system inside out. Sometimes, that only comes from getting your hands dirty in the initial architecture.
2. Why “AI Wrappers” Won’t Save You in Ushago
The market is currently flooded with AI “wrappers”—foreign AI engines with a slightly different skin. Vincent warns that these are largely useless without local contextualization.
“AI is just STATISTICS. Under the hood, it infers something,” Vincent noted. The problem? Many people take a Western solution and try to apply it to “my mom in Ushago, who doesn’t know even what a credit card is.”
To create real utility, Kenyan fintechs must prioritize data aggregation and annotation that reflects the actual behaviors of Kenyan consumers. You cannot apply a New York solution to a Kisumu problem.
3. The Rise of the “Product Engineer”
With AI and automation now handling the low-level, boilerplate code, the role of the software engineer is shifting. It is no longer enough to just write code. Engineers must become Product Engineers.
“Boilerplate stuff has been automated,” Vincent said. “For us as a team, mostly every single engineer is considered a product engineer. So spend more time with the users, try to understand how the users are navigating the system.”
This shift requires understanding social science and economics—disciplines Vincent, a non-computer science graduate, relies on daily. AI cannot understand why a customer in Nairobi uses a system differently than a customer in a rural area; only human engineers focusing on user empathy can do that.
4. Advice to Founders: Avoid Technical Debt and Look Beyond Nairobi
For non-technical founders, Vincent defines technical debt as taking shortcuts (like managing servers through a C-Panel to launch quickly) that become unscalable later. While shortcuts help with initial speed, Vincent insists that a founding technical team member is essential to “mop up” and ensure architecture is truly scalable—able to automate resource usage based on traffic peaks and valleys.
Finally, when asked about the most underrated opportunity in the Kenyan ecosystem, Vincent didn’t point to software. He pointed to hardware and energy outside of Nairobi.
He highlighted that European companies currently dominate the off-grid solar and software space in rural Kenya. “People live within the cities… People don’t care what happens let’s say, for example, in Bamuri. I think there’s that particular neglect that people have.”
