Jan Erik Solem is CEO and founder of Mapillary, a Malmö-based street-level imagery platform that that uses computer vision to fix maps. Thus far the company has won almost $25 million in funding, including from Silicon Valley powerhouse Sequoia. Solem, a computer vision specialist and former math professor, is a serial entrepreneur who exited Polar Rose, another Swedish venture, to Apple in 2010. He tells Red Herring about the “A-ha” moments that made Mapillary, competing with mapping’s huge players and the decision to build a global, distributed team right from the get-go.
When did the idea for Mapillary first arrive to you – did it feel daunting to have such a big idea for a space that’s dominated by such gigantic players, for example Google?
The idea for Mapillary came when I was working in the Bay Area, and saw first hand that even the biggest players struggle to access the street-level images and data they need to populate and update their own maps. The traditional way of updating maps involves sending out fleets of mapping vans to collect imagery, which just doesn’t scale. The idea behind Mapillary is to build a platform where anyone can use any camera to capture street-level imagery and instantly create map data in those images at scale through computer vision. It never really felt daunting to enter the space. It just made sense and we saw the need for Mapillary from day one.
What was your biggest early roadblock in turning Mapillary from a concept into a company? Was it difficult to communicate the company’s benefits to investors?
Access to map data at scale is a huge problem for everyone–from map companies that need to fix issues and update maps, to cities that need to do inventories of their street assets. Most people don’t think about this until something is broken, whether that’s a street that doesn’t appear on the map in your phone or missing or damaged traffic signs. For the organizations that spend all their time thinking about having detailed and up-to-date maps, however, access to map data is everything. So commercializing the concept was less about explaining the benefits than it was getting the technology right.
Sequoia Capital rarely invests outside the US, but they led our seed round following a conversation around our shared vision of how cameras and computer vision will become much more prominent everywhere in society over the next decades. That’s now four years ago and we’re already seeing that happen. For example, it’s estimated that there will be 25 cameras integrated into all new vehicles by 2025, and 700 trillion images will be captured daily by 2050. The data created from all those images will influence everything from how we build our cities to how our cars make decisions about transporting humans and goods.
Was there a single point which you realized the product was viable?
There wasn’t so much one single moment as there have been several “A-ha” moments over the five years we’ve been running Mapillary. We didn’t know that there would be a highly enthusiastic community of Mapillary mappers all across the globe, who run their own “mapathons” to go out and improve map coverage in their local communities. Our Japanese community even organized a Mapillary conference, with an agenda and speakers, without us instigating or even attending it.
We’re just about to hit 800 million images contributed to the platform. That people and companies would be willing to capture and share images with others, was a guess validated early. Our job is to make sure that our computer vision remains the best in the world. The data created attracts people and companies to contribute images to the platform, and because the model is collaborative, everyone gets better data as a result.
Describe the decision behind your global, and asynchronous, team–and how it works differently to other startups.
We decided to operate a distributed team from day one, and today we’re more than 50 people in 14 different countries. Operating a distributed team means that we have a global hiring pool, and that transparency and collaboration are a necessity for the team. We communicate in the open by default through public Slack channels, meaning that information is easily accessible for anyone at the company. This makes us faster and fosters great trust.
How important is it for Mapillary to scale quickly?
We are building a platform that allows anyone to access map data anywhere, and we’re always focusing on taking that to the next level. To date, you’ve had to be present at the place you want to have map data from in order to capture imagery, or hope that someone else has captured imagery at that location. We’re currently building a solution to connect supply and demand so that you can task people thousands of miles away to capture images and map data for you if you can’t go there yourself.