TOA Podcast Studio: What entrepreneurs should think about before starting their companies
Serial entrepreneur Jan Erik Solem believes that the only startup ideas worth pursuing are those that will make a real difference. That was Jan Erik’s reasoning for launching his first startup, Polar Rose, one of the first facial-recognition software companies, which was acquired by Apple in 2010. He’s now on to his second endeavor, Mapillary, a street-level imagery platform that uses computer vision to fix maps.
Solem was interviewed as part of the TOA Podcast Studio series recorded during the Tech Open Air 2019 conference. There he discussed his company’s approach to mapping data, the ethics of data use, and the lessons he’s learned as a startup founder.
Mapillary was dubbed as a rival to Google’s Street View by TechCrunch, and the Swedish startup caused a stir last year after it announced a partnership with Amazon’s Rekognition API. The API will help read the text in the photos of Mapillary’s database of 660 million crowdsourced images.
A gap in mapping data
Solem launched Mapillary six years ago after noticing the challenges startups faced collecting and sharing mapping data. He was living in Silicon Valley at the time, working at Apple after selling Polar Rose. Solem realized how difficult it was for startups and even large tech companies to obtain raw mapping data for the navigation-related apps they wanted to build.
“The reason is because all the tech companies and the bigger mapping companies, they are collecting data in a very, very expensive and unscalable way,” Solem said.
Accessing the raw mapping data and imagery is costly, so when companies acquire it they don’t want to share. Solem believed that if you opened up data collection to everyone — specifically anyone with a smartphone — you could then use computer vision to connect the crowdsourced images, extract data from them, and publish it. Doing so would eliminate information silos and open up mapping data to a whole new world of companies.
Mapillary was born to fill this data collection need. The company publishes images and mapping data that other companies use as the basis for whatever they plan to build. Mapillary purposefully doesn’t provide a mapping product, which would compete with a lot of its customers.
“If you’re horizontal and you let everyone collaborate on the data set, then you get some very interesting and nice network effects on how that data set grows,” Solem said.
But Mapillary announced earlier this year that it’s launching a marketplace to help fill in some of the gaps in its mapping data—users can make requests for images of specific areas, and individuals can fill those requests, for a fee.
Build something that matters
When students and aspiring entrepreneurs ask Solem for startup advice, one of the first things he tells them is to work on something that matters. It’s easy to underestimate how long creating and exiting a company might take, and you want to invest your time and energy into something that’s worth it.
Solem spent six years working on his first startup before it was acquired, and he’s already dedicated six years to Mapillary. “You need to think about that when you start, “Is this a problem that I can see myself working on for at least five to 10 years?”
Of course, founders can and do fail quickly. However, Solem notes that if you are successful, you’re in for a 10-year endeavor. He said entrepreneurs need to understand the time commitment from the get-go, and evaluate their ideas accordingly.
In addition, Solem stressed the importance of considering early in the process how your startup will make money. In the Web 2.0 era, the conventional wisdom held that if you built something great, revenue would follow. But Solem pushed back on that assumption, based on mistakes he made with his first company.
“As a first-time entrepreneur,” he said, “I ignored, or we ignored, the revenue side and the financial stability of the business for way too long.”
Entrepreneurs, he said, need to think about the monetization of their product from the start. Otherwise, they can find themselves in difficult situations and with an unsustainable business model.
The good – and the bad – of emerging tech
Solem has long worked at the forefront of new technologies. It’s an exciting place to be, but he contends that it also comes with responsibility.
For example, by making Mapillary’s mapping data open-source, he said the company considered the potential users and uses of their data from the beginning. “When you publish open data, when you publish open source, when you make tools available to the world, there are good and bad uses of that technology,” he said.
Solem and his team also decided early on that they would never take on any military contracts or do deal with organizations that feed into military applications. “The military, they do good things, and they do harm sometimes,” he said. “We are not the ones who can distinguish one from the other, so we say no to all of those applications.”
When Solem was developing his first company, Polar Rose, facial recognition was a nascent technology. The company was indexing faces on websites. Today, however, the technology can be used to track people — an application he didn’t anticipate. “The privacy implications are very different now in 2019 than the privacy implications in 2005,” he said.
For now, Solem is focused on helping Mapillary grow. As for what his next endeavor may be, it’s hard to say. But he’s still passionate about new applications of computer vision technology.
“The part that interests me is when you can apply it broadly, and globally in large scale,” he said. “I guess you’ll have to find an intersection between those two, and whatever the problems in the world are at the time when the next project comes along.”
Stay tuned for each new episode of the TOA Podcast Studio series on the What’s NEXT podcast feed. Co-hosted by TOA and Samsung NEXT, these sessions brought together a dozen of today’s leading minds in technology to share the human stories behind their innovations. To hear more, subscribe to What’s NEXT on Apple Podcasts, Google Play, Spotify, or via RSS on your podcast app of choice.