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 creators who value tools they can inspect, AI Company Emulator presents GewerkTon through a day-by-day simulated replay of AI agents handling a construction app’s business. It offers a close view of how an AI workforce responds to decisions and setbacks while running a real product. The replay is explicitly an emulation: customers, deals, revenue and all figures after day zero are simulated, not real GewerkTon results.

Prime Big Deal Days · Oct 6–7Offer from Amazon

Get the latest gadgets delivered free — and shop member deals

  • Fast, free delivery on millions of items
  • Access to Prime Big Deal Days deals on October 6–7
  • Prime Video, Amazon Music and more included
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.

FALL

Fall Picks

As an affiliate, we earn on qualifying purchases.

You May Also Like

The Swarm Is The Weapon: Why Agentic Attacks Break The Defensive Playbook

The rise of autonomous AI agent swarms challenges existing cybersecurity defenses by exploiting parallelism, knowledge sharing, chaining, and volume tactics.

Breaking Internal Barriers To AI Adoption

Organizations are overcoming internal resistance to AI by focusing on organizational change, partnerships, and workforce engagement, not just technology.

Why AI Art Often Raises Better Questions Than It Gives Answers

Because AI art challenges traditional notions of creativity and ownership, it leaves us questioning what truly defines art in a digital era.

The Compounding Error Problem — Why 99.9% Alignment Decays to 60% in 500 Generations

Analysis of how 99.9% alignment accuracy diminishes significantly over multiple AI generations, raising concerns for recursive self-improvement safety.