AI won’t save your tech startup if ops are broken. It’ll only accelerate your chaos — John Itodo, AI & Marketing Expert
In the global tech arena, AI is fast becoming the North Star. Investors chase the next big model. Startups pivot overnight. Chatbots get smarter. Datasets get bigger. But in the quieter corners of Africa’s tech scene, a different conversation is brewing — one that favours precision over power, context over computation.
We caught up with John Itodo, a Nigerian marketing strategist turned data advocate, turned AI and Tech Innovator for Emerging Markets, whose unusual path into AI doesn’t begin at a research lab in Silicon Valley but with a spreadsheet, a dashboard, and a stubborn refusal to accept intuition as business strategy.
Over the course of our conversation, Itodo walks us through what it means to build intelligence before infrastructure, why AI isn’t Africa’s silver bullet, why Africa’s AI moment won’t be won on scale, how investors should rethink “AI-ready” startups, and why culture eats code for breakfast.
SMEPEAKS: You’ve said before that Africa’s AI conversation shouldn’t mimic the West’s. What does that look like in practice?
John Itodo: It starts with recognising that our realities are structurally different. If you`re building in Lagos, for example, you’re not just launching a product; you’re battling patchy infrastructure, fragmented trust systems, and offline-first customers. So instead of asking, “How do we build the next OpenAI?” I ask, “What do we need today, and what lightweight intelligence can help us get there?”
That’s the mindset that shaped my entry into AI. My first brush wasn’t a neural network. It dates back to my marketing beginnings and has only crystallised through a rule-based chatbot I built in 2024 for a training institution bleeding leads after hours. No GPUs. No ML. Just a script that halved response time, boosted signups, and restored trust. That project taught me something I’ve never forgotten: intelligence doesn’t have to be artificial. It just has to be useful.
(He chuckles.) I always say, you don’t need a trillion parameters to answer a WhatsApp message at midnight.
SMEPEAKS: Your career has zig-zagged through some of Africa’s most operationally intense industries, across payments, retail, and digital training academies. Has that shaped your AI lens?
John Itodo: Completely. At Fast Pay Nigeria, where I worked on platform adoption for KePay, we were onboarding hundreds of merchants. But we weren’t learning from them. Once they joined, their data wasn’t really tracked or used. I pitched a simple feedback mechanism, built using SQL and structured forms. That alone improved merchant retention by over 15% in one quarter.
Later at LONTOR Hi-Tech, we were running campaigns across both physical stores and online channels. We didn’t have fancy dashboards. We had field-level insights, store reps reporting by phone, and messy customer data. That’s where I started building performance systems on Metabase, Power BI, and process-level analytics. Our sales jumped 20% in months because we replaced guesswork with real-time inputs.
Those roles weren’t about AI. But they forced me to think in systems, to obsess over feedback loops. That mindset is what made the transition to AI so natural. They were operations-first intelligence systems that gave me the mindset to approach AI by always rooting technology in operational clarity.
SMEPEAKS: And then you formalised that journey with a Master’s in AI?
John Itodo: Yes. I wanted to sharpen the theory behind what I was doing in the trenches. In the field, you can build intuition for data workflows. But once you enter conversations around model governance, fairness, and optimisation, you need that technical grounding.
At the University of Hull, I focused my dissertation on building an AI-powered chatbot for student engagement in healthcare education. This time it wasn’t rule-based; it was trained on a domain-specific dataset, with a real deployment and human-in-the-loop oversight. Built the pipeline from scratch, deployed it in a real academic setting, and monitored feedback. I wasn’t just exploring chat interfaces. I was interrogating what it means for a system to learn responsibly in context.
That process taught me how fragile AI is without context. It wasn’t just a project — it was a lesson in responsibility, in fairness, in ethics. That balance, practical frontline experience and formal AI training has been my edge.
SMEPEAKS: Let’s talk funding. VC decks across Africa now flash “AI-driven” like it’s a badge of honour. What do you see investors missing?
John Itodo: Too often, investors are seduced by buzzwords. But an “AI startup” with broken ops is a ticking time bomb. If you don’t have clean data, disciplined documentation, and stable workflows, AI won’t save you — it will only accelerate your chaos.
The smarter bet is on startups that already practice operational intelligence. They might not have models yet, but they have systems. They measure relentlessly. They know where their funnel leaks, why customers drop, and how to build trust. That foundation is what makes them AI-ready. Not the line in the pitch deck.
WORTH A READ:
The African startup funding winter is here — and it’s not just about money
SMEPEAKS: ChatGPT and other generative AI tools have sparked excitement across the continent. But there’s also concern around copy-pasting models trained elsewhere. What’s your take?
John Itodo: Most generative AI models are trained on global internet data — Reddit threads, Wikipedia, Western news archives, you name it. When we import them wholesale into African systems, we’re also importing biases, irrelevance, and blind spots.
We’ve seen this before with chatbots that couldn’t process local slang or context. If a system fails to understand Pidgin or Yoruba phrasing, users will drop it almost immediately. The issue isn’t capability, it’s fit. Which is why I say the real question isn’t “How big is the model?” but “How trustworthy is it in our reality?”
WORTH A READ:
How Artificial Intelligence is transforming small businesses
SMEPEAKS: Localisation has been a sticking point in AI. How should African startups approach it?
John Itodo: Localisation isn’t a feature; it’s survival. If your recommendation system ignores cultural nuance, users won’t trust it. And without trust, no AI survives in our markets.
That`s why I always push for smaller, context-rich datasets and human-in-the-loop systems. A credit scoring app for traders in Aba doesn’t need Reddit training data. It needs accurate transaction patterns and merchant behaviour data from Aba. In a nutshell, relevance is more powerful; it will always beat raw scale in our markets.
SMEPEAKS: Let’s talk tools. Why do you keep returning to low-code analytics tools like Metabase and Power BI when everyone`s obsessed with ChatGPT?
John Itodo: It`s because those are the tools that actually get used. Most African startups can’t afford full-time ML engineers or expensive GPU clusters. But they can afford dashboards, structure data better and automate feedback loops. Through that, they build data maturity, track what matters, create workflows around insight, and prepare their systems for more advanced automation down the line.
Don`t get me wrong, ChatGPT is exciting, but if you’re not tracking your lead sources or understanding churn triggers, what are you really prompting it to solve? Tools like Power BI, Metabase, and even Airtable remain the stepping stones. Power BI lets you build dashboards that inform hiring, product tweaks, and even investor reports. Metabase is great for lean teams. These aren’t just stopgaps. They are the foundation for real intelligence. These tools combined make intelligence a habit, not a hype cycle.
When teams understand their data, their questions improve. And once the questions improve, AI becomes a tool, not a distraction.
(He smiles.) If your investor asks for a traction report and you send them vibes, not a dashboard, you’ve already lost.
WORTH A READ!
Beyond the hype, how ChatGPT is quietly transforming customer service and business efficiency
SMEPEAKS: Back to investments. Funding is still pouring into Africa’s tech scene, despite global slowdowns. Partech Africa recorded $6.5 billion raised in 2022 alone. How does AI fit into that funding narrative?
John Itodo: Investors are chasing growth, but they’re also chasing discipline. They’ve seen what happens when hype outpaces systems. If you look at the valuation drops of 2022 and 2023, many of those startups weren’t struggling because the tech was bad. They were struggling because the fundamentals were broken.
That’s where AI becomes tricky. A shiny AI feature can get you headlines and maybe even a cheque. But if your operations are messy, AI will only accelerate the chaos. Investors are waking up to that. They want to see operational intelligence first. That’s the real due diligence; not just your runway, but your data habits.
SMEPEAKS: You often talk about culture. You’ve said “data-first culture before AI-first.” Why does sequencing matter?
John Itodo: Because bad habits scale faster with AI. If you don’t have documentation, if your teams don’t close feedback loops, and again, if your data’s a mess, adding AI won’t solve it. It’ll just fast-track an impending meltdown.
I’ve worked with startups that spent months building AI features nobody used. Not because the features were bad, but because they weren’t rooted in the company’s actual decision flows. Culture matters. And culture starts with how teams handle data. Not just the tech team. Everyone. When data becomes culture, AI becomes a tool. Not a toy.
SMEPEAKS: What blind spots worry you most about AI adoption in African startups?
John Itodo: Two things, actually: ethics and shortcuts.
On ethics, we’re moving fast and importing models with biases we don’t understand, yet we lack regulatory guardrails. A flawed health AI tool deployed here could magnify inequality faster than ever. We need ethicists, policymakers, and local researchers in the loop, not just engineers.
If a credit scoring model underestimates a trader in Aba because the data was biased, who’s accountable? If an AI-powered hiring tool screens out candidates from underserved communities, who checks that system? We need African voices in those rooms. Not just developers, but ethicists, policymakers, marketers, and people with lived context. We need to build systems that are just, not just fast.
On shortcuts, too many founders think they can skip operations by plugging in AI. You can’t “AI” your way out of broken processes. If 70% of your transactions need manual approval, AI will just make the mess faster.
SMEPEAKS: Looking forward, what does a winning AI ecosystem in Africa look like to you?
John Itodo: One where startups treat intelligence as culture, not as code.
That means teams document processes, measure obsessively, and build data maturity before chasing models. It means VCs fund systems thinkers as much as engineers. And it means products designed for trust, not just novelty.
If we get those habits right, AI will be an amplifier, not a crutch. But if we skip them, we’ll spend millions training models that collapse under our own operational weight.
WORTH A READ:
How your business could maximise the impending Artificial Intelligence takeover
SMEPEAKS: Final thought. If we look ahead 5 years, what kind of AI ecosystem do you hope we’ve built?
John Itodo: One where our problems define our systems, not the other way around. Where AI isn’t a prestige project, but a practical layer in solving real issues — last-mile health access, SME credit, climate risk. Where datasets reflect our realities, not just our aspirations. And where startups understand that intelligence is a habit, not a headline. Because if we get that part right, the models will follow.
Why this matters
Africa’s AI story won’t be written by the biggest models or the flashiest demos. It will be written by founders who fix their decision pipelines, by startups that turn dashboards into culture, and by innovators like John Itodo who prove that a chatbot can matter more than a trillion parameters if it solves the right problem.
For investors, the takeaway is clear: fund the habits, not the hype. For founders, the message is sharper: AI won’t save your startup if your ops are broken. It will only accelerate your chaos.
Do you have an innovative business (small, mid-scale business or startup) that adds value, creates opportunities or solves bold problems? Then you`ve got a story worth telling? Shoot us an email with SUBJECT “Story Worth Telling (+Name of Business)” to [email protected]
