TL;DR
Get office and shipping supplies 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
The AI Company Emulator is publishing a day-by-day replay of an AI team operating GewerkTon, a construction-site app. In the run through Sept. 30, 2026, the simulated team wins a pilot, ships a requested feature and converts a pilot to a paid licence; these are not real customer or revenue results.
The AI Company Emulator has published a 44-day replay of an AI team operating GewerkTon, a construction-site app, culminating in a simulated pilot becoming the product’s first paid licence, as detailed in the original analysis. The run begins with GewerkTon’s actual early-stage position, but all activity and results after day 0 are emulated and do not represent real customers, deals or revenue.
The replay assigns AI agents to product, engineering, pilot success, business development and finance, with engineering staffed by two agents—an example of AI automation in a small business. Each simulated business day is presented as a sequence of decisions and actions, then recorded as a git commit. The site also provides a live activity feed, an office map showing which roles are working, and a timeline for moving between days.
According to the replay, the team won its first pilot on day 6, shipped its first requested feature on day 16, and converted a pilot into a paid licence on day 44. The day 16 release followed an engineering review that had blocked proposed work; a founder directive then unblocked it. The site reports 13 pilots won, 10 active pilots and 48 releases by day 44, with average pilot health of 67. Those figures are simulated.
The emulated run also includes setbacks: reviews reject proposed work, offers go unrecorded, and the founder issues four short directives shown in the feed. The replay is designed to expose both stalled work and the interventions that move it forward, rather than presenting only milestones.
A 44-day simulated startup replay · Sept. 30, 2026
What Happens When an AI Team Runs a Startup Day by Day?
The AI Company Emulator follows a role-based AI team operating GewerkTon, a construction-site app. The replay makes daily decisions, setbacks and founder interventions visible—with a clear boundary between the real starting point and simulated results.
The operating cast
Five business functions are represented by agents. Engineering has two agents; the replay presents each business day as decisions and actions recorded in a git commit.
Frames needs and priorities.
Reviews and ships proposed work.
Tracks pilot health and commitments.
Handles offers and opportunities.
Represents financial decisions.
Milestones—and friction
The published run shows progress alongside the work that stalled, failed review or needed a founder directive.
A simulated prospect enters a pilot. The replay does not establish a real-world customer relationship.
Engineering review blocked proposed work; a founder directive unblocked it before the first requested feature shipped.
A pilot converts within the simulation. The reported licence is not evidence of actual GewerkTon revenue.
From daily action to visible replay
The emulator exposes how work moves through the modeled company, including oversight and coordination.
Role-based agents respond to requests and priorities.
Work, prospect follow-up and pilot commitments unfold.
Each simulated day is captured as a git commit.
Feed, office map and timeline show activity over time.
Reported by day 44
These operational figures come from the emulated run. They describe the replay, not verified business performance.
Simulated pilot wins reported by the site.
Reported as active within the replay.
Reported releases and average pilot health score.
What the replay can—and cannot—show
Its value is process visibility. Its limits matter when interpreting outcomes.
Coordination under a model
Readers can follow separate roles responding to prospects, feature requests and missed commitments, and see where founder intervention changes the next step.
Real-world traction
The published account does not establish that a replayed pilot, response, offer or licence corresponds to a real transaction. It provides no evidence of actual revenue or customer growth.
A founder, a beta tester, no customers
GewerkTon is described as an app for construction-site managers, tradespeople and facility operators. Its stated real starting position was one founder and an experienced site manager testing the beta.
Model and scoring questions
The available description does not explain model selection or configuration, how simulated prospects are generated, or how management quality is scored. That limits comparisons with real startups or other runs.
Setbacks are part of the story: proposed work gets rejected, offers go unrecorded, and the feed shows four short founder directives. The replay aims to expose stalled work and the interventions that move it forward.
Key questions
Keep the simulation boundary in view as you explore the timeline.
Did GewerkTon get its first paid customer?
The replay reports a pilot converting to a paid licence on day 44. That outcome is simulated and does not confirm a real customer or payment.
What is real in the replay?
The starting state is described as real: one founder, an experienced site manager testing the beta, and no customers. Activity after day 0 is emulated.
What does the AI team do?
Agents cover product, engineering, pilot success, business development and finance. The feed also shows founder directives and simulated prospect and pilot responses.
How far does the published run go?
As of Sept. 30, 2026, the replay covers days 1–44. The site says it adds new days daily; later milestones should be treated as simulated unless the run status is clearly identified otherwise.
What the Replay Shows About AI Teams
The project makes an AI-run company’s operating process visible over time. Readers can follow how separate role-based agents respond to prospects, feature requests and missed commitments, and see when a human founder intervenes. That offers a concrete way to examine coordination, execution and oversight in a simulated startup setting.
The distinction between simulation and business performance is central. The replay starts from GewerkTon’s real status—a founder, one experienced site manager testing the beta, and no customers—but later milestones and operating figures are generated within the emulation. They may illustrate how the system behaves under its modeled conditions; they do not establish that AI agents have delivered comparable results in an actual company.
As an affiliate, we earn on qualifying purchases.
GewerkTon’s Real Starting Point
GewerkTon is described as an app for construction-site managers, tradespeople and facility operators. At day 0, it had one founder and a beta test with an experienced site manager, but no customers. The replay says this starting state is real and labels subsequent events as simulated on every screen.
The emulator is called Firmulate, which its site describes as running AI models as complete companies with crises, financial mechanics and temptations, while scoring management quality rather than conversational quality. The GewerkTon replay covers days 1 through 44 as of Sept. 30, 2026.
“Everything after day 0 is emulated.”
— AI Company Emulator
As an affiliate, we earn on qualifying purchases.
Limits of the Simulated Results
The published account does not establish whether any replayed pilot, customer response, offer or licence corresponds to a real-world transaction. The reported paid licence is simulated, and the figures do not provide evidence of actual GewerkTon revenue or customer growth.
The available description also does not specify how the emulator selects or configures its AI models, how its simulated prospects and customers are generated, or how management quality is scored. Without those details, readers cannot assess how closely the scenarios reflect real startup operations or compare results across runs.
As an affiliate, we earn on qualifying purchases.
More Days in the GewerkTon Replay
The site says it adds replay days daily, so the next development is continued publication of the emulated team’s decisions and outcomes beyond day 44. Readers can follow the timeline and activity feed at aicompanyemulator.com. Any later milestones should be read as simulated unless the site clearly identifies a change in the run’s status.
Source: Thorsten Meyer AI
software development tracking tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
Did GewerkTon get its first paid customer?
The replay reports a pilot converting to a paid licence on day 44, but that outcome is simulated. It does not confirm a real customer or payment.
What is real in the replay?
The starting state is described as real: GewerkTon had one founder, an experienced site manager testing its beta and no customers. Activity after day 0 is emulated.
What does the AI team do?
Agents cover product, engineering, pilot success, business development and finance. The replay shows their daily decisions and actions, along with founder directives and simulated responses from prospects and pilots.
How far does the published run go?
As of Sept. 30, 2026, the replay covers days 1–44. The site says it adds new days daily.
Source: Thorsten Meyer AI
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
