How Passt Aims to Simplify the Search for a Tradesperson
In a Berlin startup programme called "Gründerszene sucht die Supergründer", 20-year-old economics student Suna Marie Tamboga is building Passt, a web platform she describes as a "Doctolib for the trades." Instead of asking a household to know whether it needs an electrician, a heating installer or a plumber, Passt asks users to describe what is broken, where the problem is and how urgent it is. It then structures the request and aims to pass it to a suitable trade business.
The underlying problem is familiar in Germany: many people do not know which trade is responsible for a damaged heating system, leaking pipe or electrical fault, so they call around, send emails and wait. Trades businesses, in turn, spend time on enquiries for work they do not offer. Passt initially makes money through a fee that customers pay when a successful appointment is booked.
Tamboga is not a programmer. She built the first version of the platform in about five weeks using OpenAI Codex, describing the functions she needed, testing the generated code and correcting errors. The platform now has separate interfaces for customers and tradespeople, a login system and an admin area. It is online and the first trades businesses have registered; the next step is to convince more businesses to list and more customers to use it.
The Marketplace and AI Logic Behind Passt
Why the Doctolib framing is more than a slogan
The core idea is structured intake before matching. A directory assumes the customer knows what they are looking for; Passt collects the problem, location and urgency first and routes the request afterwards. That targets a specific information gap: a household with a broken heating system may not know whether the job belongs to a heating installer, a plumber or an electrician. Trades businesses also avoid some unsuitable enquiries. The logic is plausible, although the article offers no market data on how many German households currently struggle with this problem.
What OpenAI Codex did—and did not—change
The fact that an economics student without software training could produce customer and trade interfaces, a login and an admin area within weeks is a concrete example of AI lowering the cost of prototyping. Tamboga's loop—describe the desired function, generate code, test it and correct it—compresses work that would previously have required a developer. But the article's own observation is important: turning software into a functioning business remains the unresolved part. Codex does not create trust, vetting procedures or two-sided demand.
Where Passt still has to prove itself
Passt now faces the standard marketplace challenge. Households will use it only if enough suitable tradespeople are listed, while tradespeople will list only if the platform sends them relevant work. The article says the first tradespeople have registered, but it gives no booking volumes or revenue figures. The long-term plan to mediate only vetted providers makes sense in a trade market built on trust, yet it adds a further operational task: building and maintaining quality checks. A fee charged only on successful bookings aligns Passt's income with completed matches, but it also means the company earns nothing until both sides are convinced.
Comments 0