Why can't ordinary people use the most powerful technology in the world?

We started by shipping games. Five of them. Puzzles, drift racing, idle empires. Nothing about it looked like the future.

It was the best possible apprenticeship. Games are the most unforgiving product category on earth — nobody is contractually obliged to open one twice. You learn, very fast, that the difference between a thing people use and a thing people abandon has almost nothing to do with how clever it is.

Then we built worlds — a fashion tower for North Fashion on Decentraland that hit third on trending places, with an NFT gallery and French DJs on the roof. We also put an AI receptionist at the door. Nobody asked for it. We built it because we'd walked through our own building and noticed the problem: it was beautiful, and there was no one to talk to.

That receptionist is the origin of everything that followed. Within months we had dropped the buildings and kept the character. Virtual Me was a companion you could design, furnish a room for, and actually talk to — GPT-4 for the mind, a customisable 3D body, a voice. In the May 2023 business plan we wrote a sentence we have been rewriting ever since:

“Bridging the gap between humans and AI, making technology more personal and relatable.”

Then the region called, and the stakes changed. We built Ahmad and Nour — the faces of TII's Falcon and Noor language models — and took them to GITEX and Web Summit Lisbon. We built an AI replica of Sultan Al Neyadi, the first Arab astronaut to walk in space, running on Falcon 180B, and presented it to the office of Sheikh Khaled bin Mohammad bin Zayed Al Nahyan. We built a Jeff Bezos AI character that drew a crowd at AWS Cloud Day UAE.

We built for First Abu Dhabi Bank, Saudi National Bank, Spinneys, and Emeritus, whose learning worlds trained Unilever's people. The realities were what we delivered. The intelligence was what mattered — and we were spending most of our budget on the part of the product nobody remembered.

So we threw away the shell. If only 2% of people used AI daily, the bottleneck was never capability. It was friction — five subscriptions, a prompt you have to learn, an hour you don't have. We started building small, sharp apps that each did exactly one useful thing with no tutorial at all.

Half a million people used something we made. We had our proof: we can take a frontier model and turn it into something a stranger will pay for on a Tuesday.

In 2025 we walked away from two #1 apps. Not because they failed — because we could see the ceiling from where we stood. We took the capability we'd proven on consumers and pointed it at businesses, where the customer pays a hundred times more and stays six times longer, and where almost nobody can actually deliver.

The tools we'd built to run our own portfolio already had names: Cortex, the AI engine. Synapse, the data layer. Today the Cortex is a brain-inspired operating system — organisational knowledge, market expertise, and autonomous AI departments in one programmable layer. Our products are clients of it.

And Virtual Minds runs on it. Founders and AI departments operating at 70–80% autonomy is not a diagram in a deck. It is our Tuesday. We are our own first customer, and the proof of the product is the company.