Why this site exists

I’m Güray Uzun, a London-based founder, and this is where I keep the record. I’m rebuilding three companies — Big Ben Suite, Big Ben Invest (the trading name of Cityy Ltd) and OverseasMED — to run on AI agents, with a named human approving every irreversible decision, and this site is the tidy home for what I learn doing it.

Most of the week-to-week happens elsewhere. I publish a build log on LinkedIn and on the Big Ben Suite journey page, where every “shipped” line has to point at a real artefact — a commit, a decision record, a corrections entry — so nothing can be quietly embellished. What was missing was one calm place to step back from the weekly cadence and write about the parts worth reviewing properly. That’s what this is.

What I’ll review here

Autonomous-company experiments— what actually happens when you point AI agents at a real business with real customers, deadlines and things that break.

Agent architectures— how the drafting, the testing gates and the approval steps fit together, and where they don’t.

Human-approval gates— the boring, load-bearing part: which decisions an agent drafts, and which a person still signs. So far, the approval layer has been more of the product than the model.

All of it is first-hand, from my own operation. When I review a development in applied AI, I’m measuring it against what I’ve watched work and fail inside my companies — not against a demo.

What I won’t do

No hype. No unverifiable claims, no invented numbers, no “this changes everything.” If I can’t back it, it doesn’t go up; if I get something wrong, the correction gets logged, not deleted.

The rule is simple, and it’s the same one I hold the agents to: receipts, not claims. New notes when there’s something real to say.