Consider Maryam. She sells garri and rice in Ikorodu, Lagos. She's been banking with Springfield Microfinance for three years — loyally, consistently, and without complaint. But every time she needs to check her balance, send money, or apply for a loan, she has to close her stall, join a queue, and wait.

On a good day, that's an hour. On a bad one, it's most of her morning. Maryam isn't unusual. She's Springfield's entire customer base.

Springfield Microfinance Bank is a licensed institution serving the people most Nigerian banks overlook — market traders, artisans, motorcycle riders, and informal workers who move fast and live on margins that don't accommodate wasted time. For years, every single service the bank offered required a physical branch visit. Not because Springfield didn't care. Because there was no other way. That changed in 2025, when Springfield hired Kati.UX to design their first digital banking experience.

The problem wasn't technology. It was trust.

Building a banking app for Springfield's customers is not the same problem as building one for someone who already banks digitally. Kati wasn't designing for people comfortable with mobile finance. He was designing for people who had never used a banking app in their lives, who had heard stories about money disappearing, and who needed to feel safe before they'd touch a single button.

That distinction shaped every decision he made. The brief had three principles: speed, simplicity, and trust. But Kati understood early that trust wasn't a feature to be added. It was the foundation on which everything else had to sit. A confusing error message, an unclear confirmation screen, a transfer that went through without showing the recipient's name — any one of those moments could lose a first-time user permanently.

So before a single screen was designed, Kati mapped every flow in FigJam. Every journey a customer might take, from opening an account to disputing a transaction, laid out in one place. Not as a process for its own sake — as a way to see where trust could break before the product existed to break it.

Designing for someone who's never done this before

The mobile app had to handle everything a Springfield customer might need: opening an account, sending money, applying for a loan, paying bills, managing cards, getting help when something went wrong. Each flow presented the same underlying challenge. How do you make something feel familiar to someone for whom nothing about it is familiar?

Onboarding

Onboarding starts with a phone number and BVN. Both are standard in Nigerian banking, which means they're instantly recognisable even to first-time app users. Kati deliberately held off on security steps — face verification, passcode, transaction PIN — until after identity was confirmed. The early screens stay light. The friction is introduced gradually, only after the user already feels settled.

Every stage has progress indicators. Every error message explains exactly what went wrong and how to fix it. And when the account is finally active, the screen shows confetti. That detail matters more than it sounds. For someone who has never successfully opened a bank account on a phone before, a celebratory screen is the moment the app tells them: you did it, and you're safe. What used to take hours in a branch now takes under ten minutes.

Money transfers

Money transfers presented a different kind of trust problem. The biggest fear Kati found among Springfield's customers wasn't fraud — it was sending money to the wrong person by mistake. His solution was real-time account name validation. Before any transfer is confirmed, the recipient's verified name appears on screen. One small feature, quietly doing enormous work. It turns a moment of anxiety into a moment of confidence.

Loan applications

Loans are broken into focused steps rather than a single long form — personal details on one screen, contact information on the next, employment status after that. On a small phone with slow internet, that structure is the difference between a completed application and an abandoned one. The repayment flow mirrors the transfer experience deliberately, so users who've already sent money once don't have to learn a new pattern to repay a loan.

Help and support

Help and Support was designed around one observation: in digital banking, the moment a user can't find help is often the moment they leave and never come back. Springfield's support flow surfaces three contact options — WhatsApp, call, and email — immediately, without navigation. Disputes are tied directly to the relevant transaction inside the app, so users don't have to re-explain what happened, and support staff doesn't have to query anything manually.

Building the tools Springfield's team actually needed

The customer-facing app was half the project. The other half was an internal admin dashboard — and it solved a different problem entirely.

Before this project, Springfield's operational team depended on IT for almost everything. Updating a customer record, freezing an account, approving a loan application, sending a notification — each action required a ticket, a wait time, and a lot of back and forth. The bank couldn't scale its service quality because every action had a bottleneck built into it.

The admin dashboard removed that bottleneck. Customer management, loan tracking, role permissions, and notification sending are now centralised in one place. Staff can freeze accounts, send reminders, approve or reject applications, and update customer details without a single IT request. New admins are added through a side panel that keeps the full user list visible in the background — a small decision that means whoever is onboarding a new team member never loses their place.

The loan management module shows key metrics at the top — totals, repayments, defaults — alongside a searchable list of every active loan. Approvals and rejections are guided with confirmation prompts that capture the reason for each decision. Nothing happens by accident. Everything is documented.

What changed

Customer onboarding time dropped from hours to under ten minutes. The internal team gained tools that let them act immediately instead of waiting on IT. And Springfield now has a platform that can welcome new customers, process their transactions, manage their loans, and resolve their disputes — entirely without the branch as a bottleneck.

But the number that matters most isn't a metric. It's Maryam, checking her balance from her stall in Ikorodu before the morning rush. No queue. No closed shop. No wasted hour. That's what the design was for.

Before Axia Africa was a school, before the accreditation and the thousands of students, Olawale Samuel was just a designer who loved showing people the way. There was no grand plan behind it. He taught because it was the most natural thing he knew how to do.

As a product designer, he had knowledge that other people needed, and withholding it never made sense to him. So he shared it. Freely, consistently, and without expectation. He taught product design to anyone who wanted to learn, and he didn't charge a thing. That instinct — to show people the way before asking anything in return — is where everything began. No pitch deck, no co-founder meeting, no business model. Just one person deciding that what he knew was worth passing on.

The referral that started everything

Axia Africa exists because someone told Olawale Samuel to stop giving things away for free. He'd been teaching product design to anyone who asked, at no charge. It felt natural to him — he knew things other people needed to know, so he shared them. Then one of the people he'd taught came back with a referral. A friend wanted to learn too. And almost as an aside, they said: you should be charging for this.

He thought about it. They were right. That conversation is where Axia began. There was no investor meeting, no formal co-founder search, just a designer realising that what he'd been doing out of instinct had actual value. He brought in a co-founder, they built the structure around it, and Axia Africa was officially a school.

What followed was the part nobody talks about when they tell the success story later. The cohorts were small, sometimes 45 students. They couldn't always pay their staff. Olawale taught the first three cohorts himself, which meant the school's ceiling and his own energy were exactly the same thing. It was just the two of them holding it together and choosing, every time things got difficult, not to stop. He had a line for it: a business only fails when you stop.

The Twitter post that changed everything

The turning point for Axia didn't come from a marketing campaign or a funding round. It came from Olawale's phone. He had been building his presence on Twitter quietly alongside running the school — posting about design, about learning, about what he was building. The audience was growing but slowly, the way most things grow when you're doing everything at once.

Then something shifted. In the space of three days, he went from 600 followers to 7,000. No paid promotion, no viral gimmick. Just a founder saying something that resonated at exactly the right moment. The effect on Axia was immediate. Founder-led sales — the idea that the person behind a company is its most powerful marketing asset — stopped being a theory and became a lived reality. People weren't just discovering Axia, they were discovering Olawale first and then finding Axia because of him. That distinction mattered. Trust in a school starts with trust in the person running it.

The numbers that followed were the kind that make you read them twice. In one cohort, with no advertising spend and no paid marketing of any kind, Axia enrolled 4,300 students. Just his posts. Just the audience he had built by showing up consistently and saying things worth reading. He holds that record to this day.

What Axia actually became

By the time you enrol in Axia Africa today, you're stepping into something that looks nothing like the scrappy three-cohort operation Olawale was running himself a few years ago. The flagship course runs for five months. Classes are live, not recorded, which matters more than it sounds — it means students are learning in real time, asking questions, getting answers, building the kind of familiarity with a subject that passive video consumption rarely produces.

At the end of every month there are exams, and those exams feed into something with real weight behind them. Axia is accredited by the American Council of Training and Development, recognised in the United States and 27 other countries. The certificate a student walks away with after five months is not just a PDF. It is a credential that holds up internationally. There is also an AI mentor built into the student dashboard, helping learners summarise videos, prepare for exams, and navigate the curriculum at their own pace. The school has students from Nigeria, Ghana, South Africa, Kenya, the UK, and beyond. The waitlist for the next cohort currently sits at around a thousand people.

But the thing Olawale is most proud of isn't the enrolment numbers or the accreditation. It's what happens after graduation. Axia graduates move into an alumni studio. They build things. They run hackathons. They start companies. And the platform the alumni studio is currently developing for global use is not being built by Olawale or his team. It is being built entirely by Axia graduates. By the people he taught. "That," he said, "is actually the goal."

The goal was never just a number

Axia Africa has committed to training 4 million Africans by 2030. It is an ambitious number, the kind that invites scepticism, and Olawale doesn't shy away from the weight of it. But he's also careful about what the number actually means. Getting to 4 million quickly is one thing. Getting there in a way that actually changes what those people do next is something harder.

He's seen what shallow training produces — people who finish a course and wait for a job that may not come. That's not what Axia is building toward. The alumni studio, the hackathons, the infrastructure for graduates to become founders rather than just employees — all of it is pointed at the same thing. Not just people who know how to design or analyse data, but people who use that knowledge to build things that didn't exist before. Tech, in Olawale's view, was never just about employability. It was always about innovation. The goal was always to produce people who create jobs, not just fill them.

He started by teaching one person for free. No plan, no business model, just the instinct that what he knew was worth passing on. That instinct built a school that has trained tens of thousands of people across the continent, holds records for the largest bootcamp in Africa, and is now being expanded by the very graduates it produced. A business only fails when you stop. Olawale Samuel has not stopped.

Most people look at a failed startup and see a cautionary tale. Oscar Hong looks at the same thing and sees a timing problem. That distinction sounds small, but it isn't.

Consider Loopt. Sam Altman co-founded it in 2005 as a location-sharing app that let you see where your friends were on a map. It raised millions, got attention, and quietly died in 2012 when Green Dot acquired it for parts. Most people filed it under "ideas that didn't work." Five years later, Snap Maps launched the same idea and holds three hundred million users. The idea worked fine. It was just early.

Oscar has been thinking about this pattern for years: the relationship between ideas and the moments they're ready to exist. He's a history nerd who also happens to build software, and that combination gave him a lens most founders don't have. History is full of ideas that arrived before the world could hold them. The question he kept coming back to wasn't "why do startups fail?" It was something more interesting: "what would happen if we went back for the ones that deserved better?" Startups.rip is his answer.

Every idea has its moment

Oscar's framework starts with a simple observation: ideas don't fail because they're wrong. They fail because the world isn't ready for them yet. He traces this back to the history of artificial intelligence. Computer scientists were proposing something like ChatGPT as far back as the 1960s. The first chatbot, ELIZA, was built in 1964 — a program that could simulate conversation well enough to fool people into thinking they were talking to a human.

For the next five decades, researchers kept pushing the idea forward. And for five decades, it kept running into the same wall. Not because the concept was flawed, but because the hardware needed to support it — specifically the GPU infrastructure that makes large language models possible — simply didn't exist yet. Then the 2020s arrived, the infrastructure caught up, and ChatGPT became the fastest growing consumer product in history.

Oscar points to Quibi as a more recent example. The short-form video platform raised $1.75 billion in venture capital before collapsing spectacularly, and the popular takeaway was that people don't want to watch serialised content on their phones. That takeaway was wrong. Apps like ReelShort, FlickReels, and DramaBox proved it. Micro-drama content built for mobile screens now generates hundreds of millions in revenue. Quibi wasn't a bad idea. It was an early one, built before the audience behaviour and creator ecosystem had fully developed around it.

Building the thing he wished existed

For a long time, Oscar tweeted about this pattern. He'd post threads about startup history, about ideas that arrived too early, about the relationship between timing and success. The response was always warm but contained — the kind of engagement you get when you're talking to people who already agree with you. Then vibecoding changed the calculus.

Vibecoding — building software by describing what you want in plain language and letting AI write the code — shifted something fundamental about who gets to build products. The bottleneck to creating software used to be specific technical knowledge. Now the bottleneck is imagination and clarity. If you can describe what you want precisely enough, you can build it. Oscar saw the opening. He'd been sitting on this idea for years: a database of failed startups paired with rebuild plans for 2026, a resource for the new generation of builders who could now actually act on what they found there.

So he built it. The entire application — all 17,000 lines of code — was assembled using AI tools without Oscar writing a single line by hand. What would have taken a team of engineers months to build came together in a fraction of that time. He draws a parallel to what happened to music when the internet arrived. You used to need a record label and a costly professional studio to put anything out. Then a MacBook and SoundCloud changed that forever. The same thing, Oscar believes, is happening to software right now. And startups.rip was built as proof of that.

The one that shouldn't have died

When you spend enough time inside a graveyard of failed startups, certain stories stay with you longer than others. For Oscar, that story is Squire.ai. The company went through Y Combinator in 2021. Its founder, Brandon Waselnuk, is Canadian — from Vancouver, the same country Oscar calls home. And what Squire built was, in hindsight, extraordinary: a proto-coding agent that automated software engineering tasks before anyone was calling it that. Their product was literally named Codex. The same name OpenAI would later give to the technology that became GitHub Copilot and eventually Claude Code.

Squire was approximately four years too early. Claude Code launched in 2025 and reached a $2.5 billion revenue run rate in nine months. The market Squire was trying to build didn't just exist — it became one of the fastest growing product categories in software history. Brandon Waselnuk had the right idea at the wrong moment. Oscar thinks about that a lot. "Imagine if the breakout coding agent was invented in Canada," he said. There's genuine feeling behind that sentence — not just admiration for a founder who was ahead of his time, but a kind of grief for what almost was.

250,000 strangers who got it

Oscar built startups.rip expecting a niche audience. Startup history nerds, mostly. People who would appreciate the database for the same reason he built it — a genuine fascination with the lifecycle of ideas and the founders who chase them. He figured that was a small group. He was wrong. Within weeks of launch, startups.rip had 250,000 users from 130 countries. Twenty-two of those countries sent at least 1,000 users each.

The diversity wasn't just geographic. The people showing up weren't all startup history nerds. Vibecoding founders were using the database to validate ideas before building them. VCs were using it to pressure-test pitches. Founders applying to Y Combinator and other accelerators were using the rebuild plans to sharpen their applications. But the detail that stayed with him most wasn't a number. It was the founders of the startups he'd profiled reaching out to say thank you — people who had built something, watched it fail, and moved on, sometimes years ago. And here was a stranger in Canada who thought their idea was worth revisiting. That response, more than the traffic numbers, told Oscar the project had found something real.

The next question

When anyone can build anything, the bottleneck shifts. It's no longer about execution. It's about knowing what to build in the first place. The question changes from "how do I build this?" to "what should I build?" And that question, it turns out, is much harder. It requires understanding which ideas have already been tried, why they didn't work at the time, and whether the conditions that caused them to fail have since changed. That's the gap startups.rip fills.

The roadmap reflects that ambition. A Pro plan already gives serious builders full database access, technical specifications, and on-demand report generation. An AI agent feature is coming. And beyond that, Oscar sees a future where the platform doesn't just provide the playbook for rebuilding a failed startup — it offers to build it for you entirely, one-click, batteries included. He built startups.rip because he wanted it to exist. Now 250,000 people across 130 countries are telling him they wanted it too. The graveyard, it turns out, was always full of the wrong kind of endings.