What was the scope of their involvement?
Eidos.ai developed the whole thing. We have an external machine learning consultant that helps us understand what Eidos.ai is doing. They developed the entire model and took care of testing and deployment.
The machine learning model preselects 30–50 relevant events in a game, such as goals and saves, and puts them in a list, so that the editors only have to watch the events selected by the machine.
What is the team composition?
We started with three people on Eidos’s side, and I believe they’ve hired two more. Gonzalo (Co-Founder) is the project director on their side, Marcelo (Co-Founder) is the senior machine learning specialist, and they have a junior machine learning specialist.
How did you come to work with Eidos.ai Machine Learning?
We found them via Upwork. We talked to multiple people, and we chose Eidos because they’re Spanish speakers, which was a big plus. Also, they’re based in Uruguay, and people there love football. They understood what the problems were going to be, and they made some proposals proactively. They proposed things that had relevance and made sense.
How much have you invested with them?
I had control over the budget during the MVP phase, but it’s gone over to our CTO now. I believe the total cost is in the neighborhood of $50,000.
What is the status of this engagement?
We started working with Eidos.ai in February 2020, and the work is ongoing. We’re working with them on automating an entirely different process now. We’ve tackled the highlights part with great results, and we’re tackling the editing of full matches now. We want to make sure that people aren’t cutting the relevant parts of the matches.