Simulated replay of the AI Company Emulator at day 44: dashboard with pilots, MRR, simulated budget and releases, the office map of emulated AI employees and the live decision feed
AIThis post was created with the assistance of artificial intelligence (AI).

For readers curious about what AI can do beyond familiar gadgets, the AI Company Emulator offers a public replay of GewerkTon, a construction-site app for site managers, tradespeople and facility operators. The run follows an emulated AI team making business decisions around a real product, with customers, deals, revenue and figures after day zero clearly simulated—not real GewerkTon results. It’s a way to see an AI workforce in action while keeping the boundary between reality and simulation clear.

Before you orderOffer from Amazon

Get the latest gadgets delivered free with Prime

  • Fast, free delivery on millions of items
  • Prime Video, Amazon Music and more included
  • Member-only deals all year
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

How it works: Each role in the company is its own AI agent — product, engineering, pilot success, business development and finance. The world around them is simulated too: prospects answer (or don’t), pilots send feature requests with deadlines, and customers get impatient when promises slip. The team notices, decides, acts and learns; every business day is a git commit that the replay walks through. A live feed shows each decision, an office map shows who is working, and a timeline lets you jump to any day.

Simulated replay running from day 38 to day 44: the counters change while the emulated AI employees work and post decisions in the live feed
The replay in motion: days 38 to 44 of the emulated run. Everything after day 0 is simulated.

The starting point is real: the run starts from GewerkTon’s actual state — one founder, an experienced site manager testing the app in beta, and no customers. Everything after day 0 is emulated, and the page says so on every screen.

What the replay shows: in the current run (days 1–44, as of 30 September 2026) the team wins its first pilot on day 6, ships its first requested feature on day 16, and on day 44 turns a pilot into the first paid licence. Along the way it gets stuck — reviews that reject everything, offers that never get recorded — and the founder steps in with short directives that you can read in the feed. That is the point: you see not just the wins but where an AI team stalls and what unblocks it.

Diagram of the simulated replay: GewerkTon's real starting state goes into the AI Company Emulator; simulated releases, pilots and a live feed come out
What goes in is GewerkTon’s real starting state; everything that comes out is simulated.

What it isn’t: a customer case study. The companies, deals and figures after day 0 are simulated, not real GewerkTon traction — the replay is an honest look at how an AI workforce behaves when it runs a real product’s business.

The emulator behind it is Firmulate (firmulate.com), which runs AI models as complete companies — with crises, money mechanics and temptations — and scores management quality, not chat quality. Watch the GewerkTon replay at aicompanyemulator.com.

HALLOWEEN

Halloween Picks

As an affiliate, we earn on qualifying purchases.

You May Also Like

What Cloud Networks Can Teach Us About AI Connectivity

Exploring how lessons from cloud computing reveal the future structure of AI infrastructure and market dynamics.

AI’s Inner Engine: Inside Twelve Machines Driving Innovation

An in-depth look at twelve key AI machines driving technological progress, explaining how they work, their significance, and what remains to be understood.

A User-Friendly App To Manage Your Badminton Results And Highlights

A new mobile app aims to simplify badminton match tracking, rankings, and highlights for local clubs and recreational players, with testing underway.

How AI Is Advancing Coding In Low-Resource Programming Languages

OpenAI has published an article about coding in low-resource programming languages, but the available information does not reveal its methods or results.