AI product studio
We design and build AI products end to end - from the first idea to the thing running in production. Thin input in, a decision-grade read out.
Selected work
Holt↗
8 language pairs · 50,932 senses · 9 AI roles under a deterministic engineWe replaced the level with a map. Holt keeps a live model of what you actually know, and assembles every session from that model - not from a lesson plan you share with everyone else.
Design + Build - strategy, interface, learning-science engineering, content pipeline · Next.js 16.2 · React 19 · Expo SDK 57 · Supabase / Deno edge functions · multi-model LLM
Live ↗ holt.iloblique.comRE Intelligence↗
Full Dubai DLD registry · 18 role-tuned modules · refreshed dailyWe took Dubai's raw public land registry and turned it into 18 focused modules - one platform where a CEO, a broker, and an analyst each see only what their job actually needs.
Design + Build - strategy, interface, data engineering, automation · Next.js 16.1 · React 19 · Supabase / PostGIS · n8n workflows · multi-model LLM
Live ↗ reint.aeHow we work
Off the obvious axis
We take the work most studios pass on - unfamiliar territory, problems without an obvious shape. The interesting answer is usually off the main axis. That's where the name comes from, and where we prefer to work.
Thin input, decision-grade out
The throughline across everything we build: take thin input and return something you can actually decide on. A few numbers in, a decision-grade read out - and the product gives you the ground to decide. It never makes the call for you.
Designed and built together
We design it and we build it - strategy, interface, and the engineering underneath, accountable to each other rather than thrown over a wall. From the first idea to the thing running in production.
About
I'm the founder of . I personally oversee every project, from the first conversation to launch, and the result is on me. The team does the work, and you always know who to hold to account.
You never see the first version. Before we show anything, we rebuild it several times: test it, cut what isn't needed, finish it properly. A site, an email and a product go through the same review.
Requests from industries we haven't worked in don't put us off. The studio has the depth to get to grips with a complex domain. First we find out how decisions are actually made there, and only then do we build.
We use AI, and we don't hide it. The value isn't in typing a prompt into a chatbot. It's in knowing where the tool can be trusted, where it has to be checked, and how to build it into the work so the result holds up in real use.
About me: twenty years in IT. I've designed the architecture and led the development of systems that serve real users under real load. That's the experience I bring to every project at the studio: seeing in advance where a system will break, and planning for it before it goes live.
If your problem doesn't fit in one sentence yet, that's a good reason to get in touch.
-- Ilia