Hi there,
Welcome to the 62nd edition of Next Capital, where we help you find Africa’s most promising startups before they get big.
In our last edition, we wrote about the startup building the Tinder for bills. If you missed it, you could catch up here 👇🏾
Next Capital is launching a podcast, and we need your brain for it.
We want you to send us your ideas.
The thing you’ve been sitting on. The gap you spotted in a market. The problem you’ve experienced one too many times and thought, “Why hasn’t anyone built this yet?”
Send it in, and we might review it on the pod.
And here’s a little secret: we’re creating something that’s going to let you build your ideas in real life.
Drop your idea here → Next Ideas Pod!
P.S. - Know someone who always has startup ideas? Forward this email to them!
Startup: Infrared
Location: Global
Documents are the backbone of modern society. Every key thing that has shaped humanity in the last 300 years started in a document - the American Declaration of Independence, the World Wide Web, Bitcoin White Paper, LLMs, just name it.
When you zoom out of the macros and in on the micros, they’re still as important today as they were 300 years ago. We all have passports, make CVs, and keep our medical histories in documents.
Roughly 13 billion documents are created every day. That’s 5 trillion a year. We only have 8 billion humans, or 75% of that if we’re counting the number of people who can actually read.
Today’s documents might not be settling disputes or creating countries, but many of them are still important. They detail IPOs, represent shares of companies, and sometimes, a person’s credit score. There’s just one issue - we can’t look at all of them as closely as we should.
This creates a host of problems.
Information overload. Too many documents equals too much data, and too much data can cloud good judgement.
Errors. Data entry error rates are around 1% and 4%. This margin could mean loss of life or bankruptcy, depending on the context.
Bad experiences. In financial services, where extensive documentation is crucial to onboard customers, 70% of companies lost clients to slow processes.
Forgery. Around 35% of documents submitted to neobanks for verification in 2024 were fraudulent. Generative AI will only make this worse.
Allen Akinkunle is familiar with all these problems. He worked for 10+ years as a data scientist and ML engineer for some of the UK’s biggest firms, like EY and Zopa Bank, classifying unstructured invoices for clients, and building decision-making tools out of them.
His day-to-day life can be put in this one-liner: everyone is drowning in documents, but no one is really reading them through. Last year, he built a tool to make everyone use documents to make faster and smarter decisions with documents.
It’s called InfraRed.
InfraRed makes sure that every document accepted from external parties gets checked and verified at signup or upon receipt, so the bad ones never get in.
It works in three ways.
First, it checks that the document is what it claims to be. If a customer uploads a selfie where you asked for a driver’s licence, InfraRed flags it in 4 seconds, so you can nudge them to try again
Second, it extracts everything. Names, dates, addresses, account numbers, all pulled from the document automatically. In a live demo, Allen ran a proof-of-address so blurry it hurt to look at (someone had scanned it with their phone, badly). InfraRed read the whole thing in under four seconds.
Third, it follows your rules. Let’s say your compliance team won’t accept a proof of address older than three months. You set that as a validation parameter, and InfraRed automatically rejects anything stale, and tells the customer exactly why.
For lenders whose business depends on finding out if borrowers are who they say they are, this is gold. In 2024, Smile ID, an identity verification startup that’s done over 110 million ID checks, found that 25% of all KYC checks that year flagged attempted fraud.
InfraRed is making it simpler to check all these docs to see if they’re legit.
Get in front of 11,000 investors, founders, execs, and tech enthusiasts in Africa.
InfraRed launched in February 2025 and has already processed over 40,000 documents.
It has had paying customers from the start, mostly financial services companies: banks, microfinance banks, and one of Africa’s largest fintechs, which uses InfraRed daily at serious volume.
Pricing is per document processed, allowing fintechs to scale the cost of this with their own user growth.
Besides directly selling to fintechs, InfraRed is going the partnership route too. Other KYC companies have started wrapping InfraRed’s API inside their products, with revenue share deals attached.
The team is made up of four people, including Allen who’s CEO and his co-founder, Ore, who spent her career at KPMG investigating financial crimes. She now runs sales and growth.
McKinsey thinks fintech revenues in Africa could reach $47 billion by 2028. That’s a five-fold increase from its 2023 estimate of $10 billion. A big portion of this will be driven by new entrants into the fintech ecosystem.These new users will all come with documents, and they’ll all carry similar risks to the ones fintechs see today.
In that world, a product like InfraRed becomes salient.
Intelligent document processing, the official space InfraRed plays in, is going to grow 6x between 2024 and 2030, hitting $12.4 billion. That’s a 33% annual growth rate.
The obvious question is, why can’t the big labs (OpenAI and Anthropic) do this? The answer is that they can do it…badly. General models are slow and expensive for jobs like this. And since accuracy matters most, small, specialized models are much better.
For InfraRed, the work goes beyond extraction. They’re adding validation layers that models don’t currently provide, and are building a slew of products to make collecting, validating, and making decisions with documents a breeze for their customers.
The future of applied AI isn’t one giant model doing everything; it’s thousands of small, fast, specialised models doing one thing brilliantly, with all the real-world context to make it superb. InfraRed is a bet on that future.
Next Capital’s take: we like founders who prove things before they ask for money. Allen ran InfraRed on his own savings for a year, launched, got paying customers, and processed 40,000 documents before raising a modest round at a sensible price. The technology demonstrably works. He demoed it live, on the ugliest documents customers can throw at him, and it works great. The wedge (document intelligence at onboarding) is narrow, but the surface it opens up (becoming the trust layer for every document moving through finance) is enormous. Disciplined founder, working product, real revenue, honest valuation. We like this one. Do you?
Want to be featured in Next Capital? Drop us a line.
Until next week! 🫡
We’re always coordinating participation across active opportunities within the Next Capital pipeline, particularly for investors who can back startups.
If you’d like to be considered for current or upcoming opportunities, let us know, and we’ll follow up where there’s alignment.
Thanks for reading Next Capital! We think more people should read it, and more people will if you share it.
Enjoyed reading this? Forward it to someone you think would enjoy it as well.
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.