What is a Fixathon?
Several teams build solutions to one company's real operational problem over 48 hours, using that company's own data — so you choose between working prototypes, not proposals.
A Fixathon is a competition where several teams build solutions to one company’s real operational problem over 48 hours, using that company’s own data. At the end, the company has working prototypes to choose between instead of proposals to read.
It looks like a hackathon, and some people call it an AI hackathon for industry. The difference is the preparation. Each challenge is scoped in advance so it can realistically be finished as a Proof of Concept (POC) in the time available, each brief ships with real operational data, and the success criteria are agreed before anyone starts coding.
The first edition ran in Porto: 48 hours, 12 technical teams, two live industrial challenges.
When it fits
If what you’re evaluating already exists as a product, run a trial or a pilot. That’s how those instruments work precisely because the product already exists.
A Fixathon is for the harder case: the solution has to be built around your context, so there’s nothing to trial yet. The usual fallback is choosing a supplier from written proposals. Here you choose after watching several of them build.
How it works
Scoping. Well before the event, we sit down with the Challenge Owner and turn a messy operational problem into a 48-hour brief: one clear question, real data, measurable criteria. Most of the value is made here, and it’s the step most innovation events skip.
Building. Pre-selected engineers, data scientists and product builders work independently against the same brief and the same data. Each team presents a working prototype at the end, validated on the real data, with a credible path to deployment.
Judging. No live demo, or no use of the real data, means no consideration. Past that gate, teams are scored on how well the solution works, its operational impact, how realistically it can be deployed, and how clearly it’s presented. An independent jury awards €10,000 to the strongest solution.
Afterwards. The Challenge Owner gets first right to adopt the winning solution, before the team takes it elsewhere.
What the first edition looked like
Blufab (Casais Group) manufactures modular bathroom panels and estimated production times from manual references and tacit shop-floor knowledge. When those estimates drifted, quotes and production planning drifted with them. Their challenge: read production time from overhead factory video and feed real estimates into the cost configurator they use to price projects. Several teams delivered working solutions; the team Cadence won the jury prize.
IEM4.0 brought a four-part operational challenge: make-or-buy classification, machine selection, raw-material list generation, and time-and-cost estimation with deviation tracking — manual, fragmented decisions that AI could make faster and more consistently.
One edition is a small sample, and we’d rather say so than oversell it. What we can say is that both Challenge Owners finished the 48h choosing among at least three credible options, and that both later opened talks with multiple Fixer teams about working together.
Who it’s for
Challenge Owners are companies with an operational problem where the deliverable is software, an application or a model built for them.
Fixer teams are the builders who compete to solve it, and who get access to real industrial data, real exposure, and direct contact with decision-makers.
Go deeper
- Buy by building — the Fixathon as an accelerated procurement process for software that has to be built.
- Fit before scale — a COO’s view on why running software on real data with real operators is the best test of whether it gets adopted.