The moment OneHazel stopped being an idea

I was sitting by the sea in Thailand. Chao Phraya River, warm evening, two CEOs from Southeast Asian gaming companies across the table from me. Different companies, different markets, but the same expression — the one that tells you someone has been circling the same problem for too long without anyone naming it.

We talked for three hours. About integration pain. About the engineering overhead of managing dozens of vendor connections. About the compliance cost of fragmented data. About how every quarter, their best engineers were spending months building pipes instead of product.

Why hasn't AI solved this yet?

One of them said it almost as an aside, more to himself than to me. But the question landed hard. Because the answer was obvious: it had not been solved yet because most of the people building AI tools for iGaming were building for the easy problems. Nobody had gone after the messy, structural, deeply embedded problem at the centre of how the industry actually operates.

The decision point

I did not write a business plan on the flight home. I did not build a financial model or run a market sizing exercise. I thought about the problem and I thought about whether I was the right person to go and solve it.

I had spent years watching operators struggle with this from the inside. I understood the commercial dynamics. I understood the compliance constraints. I understood why the existing solutions were inadequate — not because the technology was not there, but because nobody had designed for the regulatory reality of iGaming from the start.

What OneHazel actually is

OneHazel is an AI-powered integration layer for iGaming operators. One API, 300+ live connectors, 22 workflow node types. It replaces the fragmented stack of custom connectors, legacy integrations, and vendor lock-in that most mid-market operators are carrying — and it does it without the integration debt that typically comes with switching.

Six months after that conversation in Thailand, the product is live, the founding customer programme is open, and the problem is exactly as real as those two CEOs described it. I am glad I said six months and meant it.