Ideas become products.

It starts on paper.

Ajdin Okanovic.

Ideas become products.

I'm Ajdin Okanovic, IT consultant and app & game developer in Graz. 100+ products so far — two of them are live in the app stores: Žandari and Sprenity.

  • 100+products so far
  • 2free apps on iOS + Android
  • Hours or daysfrom idea to prototype

The idea

The card game my Deda taught me — on every phone.

The product · Žandari · Free on iOS + Android

My Deda taught me this game.

For me, Žandari is more than a game.

It's a card game from the Balkans, and I learned it at the table from my Deda — that's Bosnian for grandpa. So I built it the way it deserves: the real rules, on phones and tablets, so generations can keep playing together. Free, and it runs entirely on the device — no signal needed.

The real logo and cards from the game.

The idea

Relax fast. One breathing technique, nothing else.

The product · Sprenity · Free on iOS + Android

Calm down. Fast.

One breathing technique, under a minute.

Sprenity has one job: helping you relax quickly through breathing. In for 4, hold for 7, out for 8 — with nature sounds and calm affirmations if you like. Free, no account, no tracking, works offline.

  1. 4in
  2. 7hold
  3. 8out
Try it free in the app ↗

Free on iOS + Android.

The ideas

And then a hundred more — for every niche I met.

100+products so far. Two are public.

Žandari and Sprenity are out in the stores — both free. Everything else stays under wraps for now — very specific tools for very specific jobs. I show them on request, in person or on a call.

2 public · 100+ built · the rest on request

The idea

Let agents do the work. Keep a human in the loop.

This AI system I built lets me prototype fast.

Within hours or days I can build a prototype and present it. The system comes with me wherever I work: a harness, MCP servers and local AI apps — local and cloud based, built around data protection and guardrails.

  1. IdeaYou tell me the problem.
  2. Hours or daysThe system and I build a prototype.
  3. PrototypeI present it. You see it work.

Harness

The loop around the models: tools, memory, permissions, checks. It lets agents do real work — and stops them where a person has to decide.

MCP

Model Context Protocol servers connect the agents to tools and data — with clear limits on what they may read and what they may touch.

Local AI apps

Models that run on my own machines and help with the daily work. What is sensitive stays there.

Agents do the work. A human stays in the loop.

Everything runs in cycles: plan, act, check — then a human decides. If it isn't right, the feedback goes back in. If it is, it ships, and the next cycle extends it. Short cycles are why I can ship fast.

  • PlanThe idea is cut into steps.
  • ActAgents build — locally and in the cloud.
  • CheckGuardrails and checks run by themselves.
  • HumanA person decides. Nothing passes without one.
  • RedoNot right yet? The feedback goes back in.
  • ShipApproved work goes out — fast.
  • ExtendThe next cycle builds on the last one.

Simplified. The real process is more complex — this only illustrates it.

Not my field? I know people.

If I can't help you, someone in my network probably can: developers, designers and specialists around the world whose work I know and trust. Tell me what you need and I'll connect you.

Let's connect →

The next idea

This note is still empty. It could be yours.

Got an idea that wants to become a product?

Tell me what's annoying you. If I can build it, I'll tell you how — and show you a prototype within days. If I can't, I'll say so, and probably know someone who can.

hello@aokanovic.eu
LinkedIn ↗