I rarely attend startup pitch competitions anymore. They’ve grown hackneyed over the past few years, thanks to a proliferation of such events and the popularity of the TV show “Shark Tank.”
But a startup contest held Tuesday at a machine learning conference in Cambridge, MA, put a twist on the typical scenario: the startups were judged by an artificial intelligence software program, instead of a panel of investors. I took the bait and decided to check out the “Startup Battle,” part of the latest International Conference on Predictive Applications and APIs, held at Microsoft’s New England Research and Development Center.
The event was the latest sign of the growing use of data analytics and predictive modeling in the venture capital industry. As is the case for many fields, A.I. might not ever render venture capitalists obsolete—there’s still a lot of relationship-building and “qualitative” discernment involved in their jobs. But advocates of data-crunching software and machine learning argue those technologies could automate much of the time-consuming work that investors put into finding and evaluating companies.
“An expert can be outperformed by software where certain tasks are very well-structured,” says Arturo Moreno, CEO of PreSeries, the young software venture that created the A.I. startup judge. Such technologies could free investors to spend “more time on helping companies grow,” he says.
PreSeries developed software that collects reams of data about early-stage companies, such as venture funding rounds listed on Crunchbase, social media activity, and patent filings from the U.S. Patent and Trademark Office. It crunches that data to produce a score predicting each startup’s odds of getting acquired, going public, or failing, Moreno says.
For the startup competition, PreSeries used a “skill” it developed for Amazon’s voice-enabled, Internet-connected Echo speaker. A representative from each of the four competing companies—GreenSight Agronomics, Klarity, Myu.ai, and NewPearl—stood at the front of the room and answered a series of basic questions posed by the device. Those included describing the company in five words or less, sharing how many funding rounds the company has closed so far, and sharing how many years of experience the CEO has. The businesses are working in different sectors, but each says it incorporates data analytics or A.I.-related technologies in some way.
GreenSight, which uses drones and sensors to gather and analyze data for agriculture customers, scored the highest. I didn’t get to peek under the PreSeries software’s hood, but I’m guessing GreenSight won in part because it has closed the most funding rounds of the group (two) and its business ties into several sectors receiving a lot of investor interest lately (drones, analytics, agtech). (GreenSight CEO James Peverill is pictured above.)
It was the seventh A.I.-judged pitch event held by the conference series, and the second in the Boston area, Moreno says. His startup’s Alexa skill has some bugs to work out—two of the companies scored near zero, which Moreno says shouldn’t have happened.
The Alexa skill is a fun experiment, but the bigger play for PreSeries is its automated software dashboard that currently monitors more than 388,000 companies. One of the goals is to help investors discover and evaluate startups they might back, Moreno says. Investors don’t have enough time to meet with all the entrepreneurs pitching them, he says, which means they could miss out on good deals. PreSeries could help curate the best bets, Moreno says, and help investors compare the performance of their portfolio versus competing firms.
PreSeries could also reduce the amount of time entrepreneurs spend raising capital, and make the process more transparent, Moreno says.
Indeed, the company fits into a larger trend around the use of data analytics in venture capital. VC firms like GV, Correlation Ventures, and SignalFire are among those using software to aid investment decisions.
Another example is Social Capital, which this week announced a new program in which entrepreneurs from around the world can fill out a questionnaire, sharing internal metrics like revenue figures. If the data passes muster, Social Capital will invest up to $250,000—without even meeting with the company, according to TechCrunch. During a six-month trial run, Social Capital evaluated nearly 3,000 companies and agreed to invest in several dozen, the firm said in a blog post. The companies span 12 countries and various sectors. And here’s a notable outcome of the data-driven approach: 42 percent of the CEOs that the program invested in are women, and the majority of them are nonwhite, Social Capital said in the blog post.
Meanwhile, it’s still early days for PreSeries. The company spun out of BigML, a Corvallis, OR-based machine learning software firm, less than two years ago. Telefónica Open Future, an arm of the Spanish telecommunications giant aimed at supporting entrepreneurship, helped create PreSeries and is one of its shareholders, Moreno says. Wayra, a network of business accelerators and workspaces affiliated with Telefónica, is one of PreSeries’s initial customers, he adds.
Moreno was brought on this year to lead the new venture. He previously was a senior associate at Bessemer Venture Partners in Boston, and he earned his MBA from MIT’s Sloan School of Management in June. He’s now based in Madrid, Spain, and is one of three employees at PreSeries, he says.
Startups and venture capitalists aren’t known for sharing data, although there has been more transparency in the fundraising process in recent years, aided by platforms like AngelList and Crunchbase. PreSeries could play a role in those efforts, too. “We’re determined to create a virtuous effect around data and reducing volatility and risk,” Moreno says.
