[Games](https://aimantas.com/writing/games/) 6 min read Code available · bridge on mod.io and GitHub, 1 Nov I let Claude run a Transport Fever 3 company for 26 years. It's $4.2M in debt Claude has been running a Transport Fever 3 company through my AI MCP Bridge mod. Here's every run, what broke, the real numbers, and how to try the bridge yourself. **Aimantas** · Product builder 1 November 2026 Updated 2 Nov _ Screenshot Vilnius around 1923, run by Claude. Transport Fever 3, 1 Nov._ **~26** game years **-$4.2M** net worth **~$300k** bus income / yr **~160** tries on one depot [Get the bridge on the Mod Hub](https://mod.io/g/transportfever3/m/ai-mcp-bridge)[Source on GitHub](https://github.com/aimantas/tf3-mcp-bridge)[Watch part one](#watch) ## The short version - Claude plays Transport Fever 3 through a free mod and a small local server I built. - Five company runs from 1900. None made money. The best one got to about 1926 at -$4.2M. - Buses in big towns pay. Loans, wrong depots and a 0 second stop time didn't. - The bridge is free on the Mod Hub and GitHub. Setup steps are at the end. ## The score so far I gave Claude a transport company in 1900s Lithuania. Twenty six game years later it was $4.2M in debt and very proud of its buses. The buses deserve it, to be fair. By the end they earned about $300k a year. The bank wanted more than that back every year, so the company slid deeper into debt while the line list looked better and better. This is the whole story so far: how an AI plays Transport Fever 3 at all, the runs that went wrong, the depot that ate 160 tries, and the vegetable lorry that drove around empty. Everything here comes from saves, logs and the game's own Line Manager. [ [Image: Video thumbnail: This is fine. A thumbs-up next to the game's fatal error dialog and a red -$4.2M stamp] 7:34](https://www.youtube.com/watch?v=K7x_OPlIZo4) _ Video Part two: this whole story in under eight minutes. [Part one](https://www.youtube.com/watch?v=Rs6wHat6m8M) has three AIs and one crash. It loads from YouTube only when you press play._ ## How an AI plays a game it can't see The AI never touches the mouse. It talks to a mod inside the game through a small local server. 1. Claude 1. Companion server (Node.js) 1. AI MCP Bridge mod 1. Transport Fever 3 _ Diagram Commands go right, answers come back left._ Claude connects to the companion server over [MCP](https://modelcontextprotocol.io/docs/learn/architecture). The server writes small command files into the game's user folder, and the mod reads them, does the work and writes back what happened. It all goes through the game's [official modding interfaces](https://wiki.transportfever3.com/doku.php?id=modding%3Aintroduction), so the installed game stays untouched. The tools are deliberately boring. Look at the world, find towns and industries, plan a road, place a stop, make a line, buy a bus, take a loan, save. There's no free money button and no way to run any code it likes. A tycoon game where the AI can print cash proves nothing. **Rule I keep relearning:** a reply that says "done" means nothing until the game shows it. Every build gets read back from the running world, and a command with no answer counts as "unknown", never as "failed". Otherwise one lost reply buys you two depots. ## First it couldn't buy a bus The first build in early October could build exactly one road and one depot. Then it tried to buy a coach and the game said no. The test world's balance was negative, and in Transport Fever a negative balance blocks every purchase. Nobody had told the AI, and nobody had told me. The first real vehicle on a real line came on 31 October, from Claude Sonnet. It took two loans, built a road, set up two stops and a depot, bought one post coach and saved. Then it built a second road, 920 metres, which the preview said would remove a field. The game crashed in the middle of the build. The mod now refuses any build the game warns about unless the AI explicitly says it's fine. ## Four companies, four ways to lose money Every run started in 1900 with a balance of zero. None of them made money. Each one lost it differently. #### v1: borrow big A $4M loan and then a $9M one as "working capital". Five bus lines earned about $80k a year between them. Repayments were about $2.5M a year. Net worth by 1905: about -$4.8M. #### v2: stay small One loan, one town, two lines. Net worth barely moved, about -$38k over a game year. Boring, and the best result of the lot. #### v3: wrong depot It placed tram depots at $360k each instead of the $180k road depot. About $540k gone. Then a grain line that worked perfectly, into a brewery that also needs glass. The map has no glass. #### v4: empty buses Stops had a waiting time of zero, so buses left before anyone got on. It also kept buying $97k coaches with no cash cushion and ran dry. v4 did prove one useful thing. Big towns pay. Two Vilnius lines earned $107k and $174k a year with five or six coaches, while small town lines made $5k to $15k. I'd been carefully building in the places where nobody lives. ## The depot boss fight One depot join took about 160 tries and never worked. What fixed it was a tool that answers one question: where will a depot actually join? #### Before [Image: Red depot preview in Transport Fever 3 with a construction not possible warning] #### After [Image: Depot preview with its grey apron touching the street] A depot only works if its apron touches a street. A depot that looks right but floats a few metres short is useless, and the game won't tell you. Claude kept placing depots beside the street and then trying to build a little road to join them. That little road had to end in the middle of a street, which counts as a junction, and the road builder can't do that. About 160 planning calls went into one attempt. So I changed the tools instead of the instructions. `tf3_area_map` draws the streets around a town, and `tf3_depot_sites` returns places where a depot will actually join. On the next run the first depot joined on the first try. So did the second. ## The diary run: 1900 to about 1926 The goal was 1950. It got to about 1926 before the humans closed the game for the night. [Image: Vilnius from above in Transport Fever 3, about 1923, with bus stops in the streets] _ Screenshot Vilnius around 1923, two bus lines in. Transport Fever 3, 1 Nov._ This run used everything learned so far. One cheap loan, depots from the new tool, bus lines in Klaipėda first, then a 298 metre road to Kretinga that worked first time. By 1905 Vilnius had its own depot and two lines. The balance said $5.3M. The loans said $7.9M. Then came freight. A farm near Klaipėda grows about 110 units of vegetables a year, and Klaipėda wants them. Here's how that went: 1. **The platform wouldn't fit.** Every spot near the farm collided with silos or fields. 1. **It fit 180 m away.** Outside the farm's catchment, so the lorry drove back and forth with nothing in it. Sold. 1. **Right beside the farm, it loaded.** Four units per cart. Then the town stop refused to accept them, and Force Unload emptied the carts for $0. 1. **The real bug was a zero.** The unload stop's waiting time was 0 seconds, a default from my own bridge. Give it a real stop time, switch Force Unload off, and three units paid about $8k each. [Image: Transport Fever 3 Line Manager listing Klaipeda and Vilnius bus lines and a Vegetables line] _ Screenshot The Line Manager around 1925. Buses earn, the Vegetables line is still red. Transport Fever 3, 1 Nov._ Meanwhile the buses quietly carried the company. Klaipėda to Kretinga earned about $150k a year by the end, Vilnius 2 about $130k. All lines together made roughly $280k to $330k a year. The bank also kept offering more than my script was allowed to ask for, so Claude took a $3M loan and later a $5M one. Final page: about 1926, net worth about -$4.2M. ## What it learned (and what I learned) #### Big towns pay Dense town pairs earn ten times more per bus. Small towns are scenery. #### The bank isn't a strategy Every run that borrowed to grow drowned in repayments. Income per year has to beat loan service before anything else. #### Check the zeros A 0 second stop time, a 0 balance, a 0 in catchment. Most of my worst bugs were a zero nobody looked at. #### Fix the tool, not the prompt Better instructions never fixed depots. A tool that answers "where will this join?" fixed them in one run. The AI is a decent dispatcher and a terrible banker. It's careful with roads and reckless with loans, which is probably the most human thing about it. ## Play with it yourself The bridge is free. It's a prototype, it's slow at placing things, and it runs on PC, Mac and Linux. If you want to watch an AI make the same mistakes with your money, here's everything: 1. **Subscribe to the mod.** [AI MCP Bridge on the Mod Hub](https://mod.io/g/transportfever3/m/ai-mcp-bridge), or search for it in the game's Mod Hub. 1. **Get the companion server.** The [companion download](https://aimantas.com/assets/downloads/ai-mcp-bridge-companion.zip), or the source on [GitHub](https://github.com/aimantas/tf3-mcp-bridge). Its README has the setup and a check you can run. 1. **Pick a map.** I play on my [Lithuania map](https://aimantas.com/writing/lithuania-transport-fever-3-map/). Germany, Switzerland, Austria, Hungary and the Netherlands are up too. 1. **Start small.** Ask it to read the world and change nothing. Then ask for one bus line in the biggest town. Then keep an eye on the loans. **Limit:** the bridge places stops and depots through the game's own placement tool, which is the slow part. Expect minutes per depot, and save before you let it build anything. Next up is a championship: several AIs, the same save, the same rules, and a scoreboard read from the game itself. The prep goes on my [YouTube channel](https://www.youtube.com/@aimantas-com) first, because the prep is where everything breaks. Sources6 references 1. Company runs v1 to v4: game saves and run notes, 31 Oct to 1 Nov 2026. Figures read from the game's Line Manager and snapshots, not from the AI's own reports. 1. Diary run: 1 Nov 2026, start save 1900 with a balance of 0, last save 1917, game closed at about 1926. Net worth = balance about $4.7M minus loans about $8.9M. 1. Depot join: about 160 plan and preview calls on one attempt, 1 Nov 2026. After the depot-site tool: 2 depots placed, 2 joined on the next run. 1. First vehicle on a line: Claude Sonnet test run, 31 Oct 2026, ended by a game crash during a road build. 1. Public pages checked 1 Nov 2026: [AI MCP Bridge](https://mod.io/g/transportfever3/m/ai-mcp-bridge), [GitHub](https://github.com/aimantas/tf3-mcp-bridge), [part one video](https://www.youtube.com/watch?v=Rs6wHat6m8M); [part two video](https://www.youtube.com/watch?v=aj00ILo8L4w) published 2 Nov 2026, and its [re-voiced cut](https://www.youtube.com/watch?v=K7x_OPlIZo4) on 5 Nov 2026. 1. MCP architecture: [modelcontextprotocol.io](https://modelcontextprotocol.io/docs/learn/architecture). Game modding: the [official Transport Fever 3 wiki](https://wiki.transportfever3.com/doku.php?id=modding%3Aintroduction). [ 7:34**I Let Claude Run a Transport Company for 26 Years** Watch the video](#watch)[Open on YouTube ↗](https://www.youtube.com/watch?v=K7x_OPlIZo4) ## Get it - [ AI MCP BridgeMod Hub ↗](https://mod.io/g/transportfever3/m/ai-mcp-bridge) - [ Bridge sourceGitHub ↗](https://github.com/aimantas/tf3-mcp-bridge) - [ Lithuania mapArticle ↗](https://aimantas.com/writing/lithuania-transport-fever-3-map/) - [ Country mapsGitHub ↗](https://github.com/aimantas/tf3-country-maps) [ **Aimantas**@aimantasai Short updates from the builds, between articles. Follow on X](https://x.com/intent/follow?screen_name=aimantasai) [ 7:34**I Let Claude Run a Transport Company for 26 Years** Watch the video](#watch)[Open on YouTube ↗](https://www.youtube.com/watch?v=K7x_OPlIZo4) ## Get it - [ AI MCP BridgeMod Hub ↗](https://mod.io/g/transportfever3/m/ai-mcp-bridge) - [ Bridge sourceGitHub ↗](https://github.com/aimantas/tf3-mcp-bridge) - [ Lithuania mapArticle ↗](https://aimantas.com/writing/lithuania-transport-fever-3-map/) - [ Country mapsGitHub ↗](https://github.com/aimantas/tf3-country-maps) [ **Aimantas**@aimantasai Short updates from the builds, between articles. Follow on X](https://x.com/intent/follow?screen_name=aimantasai) ## In this article 1. [The score so far](#the-score) 1. [How an AI plays a game it can't see](#how-it-plays) 1. [First it couldn't buy a bus](#first-steps) 1. [Four companies, four ways to lose money](#the-runs) 1. [The depot boss fight](#depots) 1. [The diary run: 1900 to about 1926](#diary) 1. [What it learned (and what I learned)](#lessons) 1. [Play with it yourself](#try-it) 1. [Sources](#sources) ## Keep reading[All 14 articles →](https://aimantas.com/writing/) [ Newer Marketing 4 min read ### I spent 1,200 Higgsfield credits in 56 minutes. An honest review €59 bought 1,200 Higgsfield credits. Six 1080p video clips took 62.6% of them in under an hour. Here's the full breakdown, what it bought, and what I'd do differently. 7 November 2026 Read → ](https://aimantas.com/writing/higgsfield-credits-honest-review/)[ Older Games 5 min read ### I made a Lithuania map for Transport Fever 3. Here's how to make your own Free Transport Fever 3 maps of Lithuania, Germany, Switzerland, Austria, Hungary and the Netherlands, and a step by step way to make one of your own country, with prompts you can give Claude. 28 October 2026 Read → ](https://aimantas.com/writing/lithuania-transport-fever-3-map/)[ Recent Design 6 min read ### I redesigned my homepage with Claude Fable 5.2. Before and after One sentence of brief. The dashboard in my hero became a deck of seven real product screenshots, and the numbers moved into one band. Before and after screenshots, the four things that broke and what it cost in bytes. 10 November 2026 Read → ](https://aimantas.com/writing/homepage-redesign-claude-fable-5-2/)