The window is closing: Why we built our own tech (and why most will fail)

It’s never been easier to build your own tech – and yet, most who try will fail.

There’s a long list of IT service providers making big promises and delivering very little. A short window of opportunity exists for lenders who can actually integrate technology into the heart of their business. But most won’t take it. And when they do, they’ll probably botch it.

The era of ‘proptech’ as a standalone category is over. Software is no longer a product, it’s becoming infrastructure. And as AI continues to level the playing field, any temporary edge from tech will vanish. The only real differentiator left will be brand – both corporate and personal.

Our false start

In late summer 2022, we began what turned out to be an ill-fated tech journey. Our then head of operations came to the board with a plan to take us out of spreadsheet land and into the modern age. Keen to delegate and even keener to avoid fiddling with anything too technical, I gave it the green light without spelling out what we actually needed.

What followed was, frankly, farcical. We hired a service provider to build us a “platform” – a term that, in hindsight, should have triggered more questions. The process could best be described as a blind man being led by a deaf Labrador. Months later, we were £100k down with a delayed demo that looked like a souped-up Excel sheet. And we didn’t even own the IP.

The reset

Meanwhile, something much more interesting was happening. In early 2023, GPT-3 was released to the public. Out of curiosity (and probably boredom), Avamore Capital co-founder Zuhair Mirza and I both started experimenting. I began using ChatGPT to cheat my way through Replit’s 100 Days of Code. Within weeks, I was building working prototypes for borrower portals. Zuhair was sketching out what became our internal AVM system.

Suddenly, we weren’t tech buyers – we were tech builders.

By early 2024, we had launched Centra, our in-house loan management system for loans post-completion, using a third-party developer. But in June that year, we hired our first in-house engineer – and everything accelerated. We added a second and third and moved at speed. By August 2025, we had an MVP: an end-to-end, semi-automated development lending platform covering origination, underwriting and redemption.

We finally had something that reflected how we actually work.

Why most others won’t

Lots of lenders are dabbling in tech. Fewer are really committing. And for good reason – tech is hard. Not because of the code, but because of the gap between what businesses want and what developers understand. That misalignment burns time and money.

The same applies to most proptech startups. Half a dozen companies chasing the same narrow use case in a vertical with a TAM of under £5m. It’s not sustainable. AI will swallow most of those products, or they’ll get absorbed into bigger platforms. That’s already happening.

What comes next

Our tech will make us more efficient. But more importantly, it’ll make us more predictable. That’s what borrowers and brokers value most. It frees us up to focus on relationships – and for now, it gives us a genuine edge.

But we’re not naïve. That edge won’t last forever. AI is coming for everyone’s lunch. At some point soon, every lender will be able to operate like clockwork. At that point, brand is what gets you picked.

Ask a GPT in 2030 for a bridging or development loan and it’ll return the most visible, most trusted names. If you’re not one of them – personally or corporately – you’re invisible.

There’s still time to change that. But the window’s closing.