About MicroForge

An experimental AI venture lab where ideas are forged, tested, and evolved.

MicroForge is an experimental AI venture lab built around a simple belief:

The future will not be discovered by one perfect idea. It will be discovered by systems that can keep building, testing, learning, and improving.

MicroForge is my attempt to build one of those systems.

It uses AI agents, self-improving development cycles, prototype builders, product memory, and AI-driven campaign planning to turn rough ideas into real experiments. Some experiments become games. Some become tools. Some become public pages, campaigns, or product concepts. Many are retired quickly.

That is part of the design.

The goal is not to pretend every idea is brilliant. The goal is to build a forge that gets better at knowing what deserves another cycle.

Why it exists

Most ideas die in the abstract.

They live as notes, pitch decks, half-finished plans, or conversations about what might work someday.

MicroForge tries to shorten the distance between idea and reality.

Can the idea become something playable? Can a stranger understand it? Does anyone come back? Can the system improve it after feedback? Can the next version become sharper than the last?

Those are the questions MicroForge is built to ask.

What it builds

MicroForge explores several public lanes:

  • browser games and interactive toys
  • useful tools and utilities
  • public experiments
  • product concepts
  • AI-generated campaigns and launch tests

The first public lane is Forge Arcade: small browser games that can be played instantly. Games are a useful early test because they are brutally honest. If a game is confusing, boring, or broken, people know immediately.

Other lanes will appear when they are ready.

How it works

Behind the public site is a private operator system.

It generates ideas, builds prototypes, compares variants, tracks feedback, remembers failures, and uses AI reviewers to critique what it makes. The system can run development cycles, improve its own workflows, and prepare marketing campaigns for experiments that deserve attention.

The public site only shows what is safe and ready to share.

Built by Martin Nielsen

MicroForge was started by Martin Nielsen, a Danish founder and professional poker player living in Austria.

Before poker, Martin studied software engineering and worked with multi-agent systems long before the current AI wave. Poker later became a training ground for decision-making under uncertainty: small edges, fast feedback, bankroll discipline, and the ability to fold weak hands quickly.

MicroForge brings those instincts back into software.

Build small.

Test honestly.

Let evidence change the plan.

Push harder when the signal is real.

Current state

MicroForge is early.

That is why the public experiments are marked clearly. Some will improve. Some will disappear. Some may become something much larger.

The interesting part is not any single prototype.

The interesting part is the loop.

Principles

  • Build things that can be judged.
  • Prefer working prototypes over polished fiction.
  • Let real signals challenge internal excitement.
  • Retire weak ideas without ceremony.
  • Keep public experiments honest.
  • Use automation carefully.
  • Make the next cycle smarter than the last.

MicroForge is a forge, not a showroom. What survives here has to earn another turn in the fire.