The AI Adviser Who Supported a Ministry to Build Their Own Tool

Anthony Udeh is an AI adviser based in Abuja, Nigeria. He has twenty years’ experience working across health, education, and agriculture development programmes, using technology to solve problems on the ground, from Ebola contact tracing, to preventing mother to child transmission of HIV, to AI-driven reading assessments. That breadth of experience, seeing how government systems work across multiple countries and sectors, now feeds directly into his current role. 

These days he advises Nigeria’s Federal Ministry of Education, where he coached the ministry’s own staff to build AskFME, which the ministry describes as Nigeria’s first AI-powered national education service desk. It lives on WhatsApp and the ministry’s website, so people reach it on an app they already have. Ask it a policy question and it answers from official ministry documents rather than inventing something. Ask it for an education statistic and it draws on the national school census.

Most of the team he coached had no technology background at all. Within four months they were writing scripts for AI agents, doing prompt engineering, and talking fluently about how documents are broken up and stored so the system can search them. His own measure of success is whether the ministry can keep building once he has gone. 

Over six months and across six countries, six ministries of education joined EdTech Hub’s Ministry of Education AI Challenge. Each Ministry worked with embedded AI experts to build their first AI prototypes. Meet the AI Experts introduces the people behind those tools: who they are, how they think, and what it takes to build AI a ministry of education can actually use. For this profile, we’re featuring Anthony Udeh from Nigeria.

Tell us about your background. What brought you to advising a government on artificial intelligence?

I started working in IT for a development agency straight after college, and went through university while I was still working there. From IT, I moved into the development sector, supporting health projects, then programmes for vulnerable children, then maternal and child health. During the 2014 Ebola crisis, I was leading a lot of the technology work, contact tracing and the systems around it, and we later took that same technology and used it to support education projects.

From there, I worked in agriculture, managing the technology side of innovation grants for companies building solutions for farmers and markets, which is where I learned how market systems actually work. Then I went back into education, which completed the cycle. I was an implementation specialist for adaptive education technology, working through COVID and travelling once it lifted, in the Philippines, Pakistan, Liberia, Senegal and a number of countries in Africa, advising ministries on technology strategy.

One piece of work I’m still proud of is an AI-powered computer-based reading assessment in the Philippines. Students did the assessment themselves, the system listened to them read aloud, and it produced recommendations. That cut down the time teachers were spending assessing every child before they could plan anything.

What does a good day in this work look like? And a hard one?

A good day is seeing work you’ve put years into being picked up by somebody, even if it’s only one percent of it. Being an innovator is one of the most difficult things to do: you sacrifice your time, and your family doesn’t really have your full attention, because your brain has to be active all the time. So when an idea is taken up, by an organisation and then eventually by the government, that’s something big.

A bad day is putting all of that in and having it go unused, or having it reduced to something small. You do all this work and the response is, thanks, but we only wanted three bullet points.

Back in 2015 I’d built and piloted an innovative education solution that worked. It was turned down in front of the funder, and I was refused the budget to take it to other local governments. I didn’t understand why at the time. Later, I realised it wasn’t really about the solution. It was going to change a lot of people’s roles, and they were afraid of what that meant for them. Years later, I walked into the ministry and found a dashboard and a mobile app, and people were building on the very pilot I had run in 2015. The people who had been nervous about it then were the ones championing it now. I wondered what might have happened if the answer had been yes back then, and how much progress we could have made together.

In plain terms, what have you been doing as the ministry’s AI adviser?

Three things at once. The first is the tool. I coached the ministry’s team to design and build AskFME rather than building it for them. It answers policy questions from a secured library of ministry documents, retrieves education statistics from the national school census, and connects people to the right part of the ministry, all without anyone needing a new app.

The second is the team. With the country lead, Abiola Sanusi, I ran a structured programme to build their skills, from working out where AI is actually useful, to preparing documents for the system, to evaluating how well it performs. They completed a formal certification through Apolitical’s Government AI Campus, and they can now manage AskFME themselves. 

The third is advice: helping the ministry choose what to build, and supporting its wider thinking on how to adopt AI responsibly.

If you were briefing someone stepping into a role like yours tomorrow, inside a federal ministry, what would you tell them?

The biggest factor in whether a technology solution succeeds in government is whether it’s aligned to what the government is already trying to do. Find the reform, the policy or the driver that your work can attach to, and then keep aligning to it, in your advice and in the way you talk about the work. Governments may have two or three years to deliver change, which is very short. If they find something that helps them move the needle, even a little, they will latch onto it and fund it. If your solution is not a political asset, something that genuinely helps them deliver, it’s dead on arrival.

The related point is money. The first thing you will be told is that there’s no additional money. Everything has to align to the budget, so if you can’t say how your idea maps onto something already in the budget, it won’t get funded. If it does map onto a budget line, that’s great, because they’re often looking for better ways to spend the budget than recycling last year’s activities.

And be careful how you talk about gaps. In a closed system, very few people will happily agree that they have a gap. So don’t just show them the gap. Show them the opportunity, and what they’d gain by filling it.

What’s the most important skill to master in your role?

You have to be good at listening. You’re listening for the unspoken things, and for who actually holds influence in the room. At the start you usually meet the people who have been sent to find out what you’re proposing. Once they see that your idea has strategic or budgetary value, you start meeting the people who make the decisions. You have to listen carefully enough to work out who really holds the key to the door you are trying to open.

You also have to understand that a lot of decisions are shaped informally. There are people around a leader whose job is effectively to carry ideas back and forth, and a great deal gets settled in conversations outside the office rather than in a meeting. Official meetings are formal and crowded, and there is often not much room in them to think. So you need to understand how the people you’re advising work, and where the real listening happens.

When you look back at AskFME, what stands out?

What stands out to me is the team and their potential. These were regular ministry staff, many of whom came from departments with no technology background at all. They sat down together to think about the problems of the Federal Ministry of Education and started building use cases. I honestly felt that was impossible. It’s not something we could have achieved ten years ago as an implementing partner.

In four months, the team went from no information about AI, to writing scripts for AI agents, prompt engineering, and understanding how documents are broken into pieces and turned into something a system can search. These are things you’d normally learn in a computer science class, where plenty of students would struggle. That was the biggest surprise for me: people can learn very fast if you translate the technical language into a format that makes sense to them and you give them the time.

And the guidance material we were given as a programme mattered, but only because it was contextualised. Guidelines are important, and then someone has to translate them into the realities of that particular country and that particular team. 

What was the hardest part of getting the team to a place where they could run it themselves?

Getting them to believe in themselves, and getting past the fear that AskFME would take their jobs. Once they believe, they start demystifying the whole thing. They understand that a lot of this is normal manual workflows being automated, that you’re giving a system instructions about how to think and what to do, and that you can adjust all of it. Then they start connecting it to their own work.

They went further than anything I asked for. They took the courses we had put them through and used them to build their own internal training programme for the whole ministry, with different packages for directors, for heads of department and for everyday staff, using the same thinking about user profiles that they had learned building AskFME. They wrote the training plan themselves.

Some parts of the ministry saw AskFME as a threat. How did you handle that?

When a solution starts to seem like a threat, it usually means it’s genuinely good. For example, one team would spend months each year cleaning data and drafting content for an education digest. AskFME produces the same work in minutes, then generates tailored versions, using the language parents (for example) actually speak, aligned with the government’s reforms, at scale.

So FME staff asked, reasonably, what that meant for them. The answer is that this is a modernisation of how the ministry serves people, and the Minister has been explicit in backing it. His argument is that there is a correlation between how well a system innovates and how much people trust it. If your education standard is really good, people should not be struggling to get information or waiting in long queues for a process. It should be easy. AskFME is one of the ways to show that.

If a ministry in another country wanted to build something like AskFME, what would you tell them before they start?

Do not assume you know what the problem is. Send out a wide survey and let people express themselves in it as much as they can. Then use that data to work out what problem your solution should actually solve. Include your own government staff in that survey, and make sure the solution solves their problems too, because it’s important to address their needs too. The more effective your staff are, the easier it’ll be for them to solve other people’s problems. 

The second thing is data. Agree it before you build. Work out who holds the information, get an agreement to access it, and settle how it will be processed into something useful for whoever is going to consume it. Without data and information, these solutions won’t work.

If you had unlimited time and funding, what would you build for Nigerian education?

I would personally want to address financial knowledge. A lot of people go through their whole lives without being taught anything about money or sustainability. That education is largely missing in lower and middle income countries. People have resources, they have land, and they don’t always know what to do with it. Knowing how to invest and how to build something that lasts becomes a burden rather than an opportunity. The curriculum in Nigeria has changed and it now includes more practical stages, but that’s mostly preparing Nigerians to be workers. What prepares them to be entrepreneurs? What prepares them to be good investors, or to manage what they build? That’s what I’d want to innovate around, an upskilling offer linked to education and genuinely accessible to a lot of people.

Anthony Udeh is the AI adviser to Nigeria’s Federal Ministry of Education in the Ministry of Education AI Challenge, an EdTech Hub programme.

Read the next interview >

< Back to all interviews