Our investment into Matroid in Q3 2020 was the culmination of nearly a 2.5-year deep dive into the types of problems that computer vision can help address in energy industrial applications.
In that process we learned that the industry subject matter experts wanted to be involved in the software training – and in fact, their participation in training the machine learning computer vision models was required. We met with about 70 companies over that time period and looked for firms that had a product these SMEs could learn and train quickly. For example, with Matroid a subject matter expert can quickly train a CV model to label what type of rust or hairline cracks in a wind turbine are critical and what are cosmetic. This used to take tens of hours of studying all drone and hi-res photographs.
That extended research process led to our investment into Matroid. (Some pages from our 60+ page research report are below.)
Matroid has stayed incredibly lean (under 20 employees) and has the financial profile of a company 5-10x the size. Why? The company is embodying product-led growth. Founder, Reza Zadeh, and his team have continuously worked to improve the product to ease integrations, lower the cost of the product and build their library of trained models.
The company’s progress was acknowledged this week when Forrester released a report on the computer vision market and “Named Matroid the Most Mature Computer Vision Platform among 43 vendors“. And some of the names on that list are very impressive.
You are likely going to be hearing a lot about Matroid in the coming year. If you want to get introduced to the company, let me know.