The automotive industry moves trillions a year and still runs on a phone queue, a web form and a salesperson who calls back tomorrow. A car is sold, financed, serviced and traded the way it was in 1914, with a website on top. We have set out to put that industry on autonomous rails: commercial operations that run themselves. AI employees that talk to every customer in seconds, in their language, on every channel, that decide, act, book straight into the systems the industry runs on, and get better every day. Dealer groups first. Then the importer, the OEM, the fleet, the finance company, every place a vehicle meets a customer.
It already runs: about a thousand real customer conversations a day for BMW, MINI, Ford and Hyundai dealer groups, and a pilot with a Mercedes-Benz group in Germany. What powers it is not a chatbot. It is a distributed system with language models as components inside it: state, queues, retries, dealer systems that answer in fifteen seconds or not at all, a hundred ways for an output to pass every check and still be wrong. For the next stage we are looking for a strong team member at co-founder level who wants to go deep on autonomous systems, model development, system design, and the real, live deployment of a new breed of system, and who takes the hardest problems because they are the hardest. Flat hierarchy: you are treated as an equal member of the team and your reward is linked to the outcome. The team pushes the frontier relentlessly. There are no meetings and no politics. There is a system in production, customers on the other end, and the next thing to ship.
The mission
Put the automotive industry on autonomous rails.
Commercial operations that run themselves. Dealer groups first, then the importer, the OEM, the fleet, the finance company: every place a vehicle meets a customer.
1,200
customer conversations a day
11,000
workshop bookings written into the DMS
4
brands live: BMW, MINI, Ford, Hyundai
1 in 4
cars sold start with the agent
Counted on production, 12.09.2026. In production, not a demo.
What you will build and do
- 01Own the agent runtime end to end: orchestration, tool execution, memory, handover to humans, integrations into dealer management systems and messaging channels, in production for real customers from your first month.
- 02Build evals and traces before features. Every change to an agent ships with a way to prove it did not make things worse.
- 03Decide the architecture with the founders: what runs where, which model does which job, what a session may cost, what happens when a component fails.
- 04Ship every week. The product is in production; the customers are dealer groups whose customers are waiting on the other end.
Who we look for
- 01Has built AI systems from the ground up, not wrappers over an API. Strong senior experience shipping systems in production. A PhD is a plus, not a requirement.
- 02Sees an agentic system as a distributed system with the LLM as one component or orchestrator inside it. Thinks in state, failure modes, retries and observability, not just prompts.
- 03Uses coding agents effectively as a daily working tool.
- 04Builds evals and traces before building features. Understands that a large share of failures here are semantic, not mechanical: output can pass every check and still be wrong.
- 05Conscious of cost and latency: how many tokens, how many dollars per session, where a smaller model does the job.
- 06Comfortable with incomplete specs. Makes decisions with missing information, documents them, and keeps moving. This is an early stage startup; nobody will tell you what to do step by step.
- 07Wants a flat hierarchy and means it: takes on the hardest problems, argues for what is right, and carries the outcome as an equal member of the team, with a reward linked to that outcome.
- 08Wants to go deep: autonomous systems, model development and fine tuning, system design, and deployments that real customers depend on the same day. No meetings, no politics, shipping.
What we offer
- Role
- Full time, remote from wherever you build best, with time together in person when it matters.
- Pay
- A fixed salary and a share in the company that links you to the outcome, discussed openly in the first conversation.
- Work
- Direct work with the founder and our AI lead, who built the current system. No meetings, no politics. A product in production with the most complete automotive AI reference in the region; your code answers real customers the week it ships.
How we work
01
Flat hierarchy
You are an equal member of the team. The best argument wins, whoever makes it.
02
Rewarded on the outcome
A share in the company links you to what we build, not to a title.
03
No meetings, no politics
Written decisions, a system in production, customers on the other end.
04
Ship every week
Your work answers real customers the week it lands. We push the frontier and measure it.
How to apply
One page: the best AI system you built from the ground up and what it did in production, how you evaluated it, and the first thing you would change about how agentic systems are usually built. A link to code we can read beats any cover letter. We answer within 48 hours. Shortlisted candidates get a paid 48 hour work sample.