VCs have been trying to crack the secret sauce to picking a winning startup for decades. But could an A.I. algorithm be just as good, or better, than human intuition when it comes to spotting the next tech unicorn?
That’s what Mattias Ljungman, the founder of Moonfire, a UK-based early-stage investing company is banking on. Today, the firm announced its second fund of $115 million. Yet unlike other early-stage funds, a proprietary A.I. algorithm the company built will be spotting many of the fintech, gaming, and health care companies that the firm invests in across Europe.
As a venture investor who helped found VC firm Atomico before starting Moonfire, Ljungman questioned whether the way VCs were traditionally sourcing founders was casting a wide enough net. “There's a lot of mental labor involved in venture,” he explained to Fortune. “The world of venture has expanded dramatically and so it’s ridiculous to think ‘Hey, I know a few people, I have a big network of people, and therefore I'm going to have the best insights to the best companies in my specific domain,’” he explained.
Computer scientist and partner at the firm Mike Arpaia started building the firm’s A.I. platform in 2020 before the firm's first $60 million fund launched in 2021. For the first year, he worked on the project alone, refining its data selection. After a year, two engineers joined the team and helped build the platform’s transformer models.
Now, up and running, the A.I.-powered platform works to source companies and help partners move the investing process along faster. The technology scours through 50,000 companies per week. Then, using its algorithm that Arpaia fine-tuned to Moonfire’s vetting system, it evaluates the startup’s founders, sector fit, and uses text that Moonfire investors themselves have written to gauge whether a startup would be a fit. “We've amassed hundreds of pages of ‘investment theses’ and our company evaluation uses this natural language text as a source of evaluation for companies,” explained Arpaia. The program then narrows down the startups to about a hundred per week and Moonfire’s human investors take a look to decide which firms are worth looking into further.
From there, the company’s technology assists investors as they vet each startup by making laborious tasks easier—like amalgamating all the information about a company in one place and putting it in the right columns, or writing an investment memo based on notes. “The machine doesn't tell you what to do, the machine helps you with the filtering process. It's a human and machine symbiotic relationship,” said Ljungman.
The firm, which has seven employees, also sources deals from traditional venture networks, however, those companies are still run through the A.I. algorithm. Ljungman explained that in Moonfire’s first fund, the split between companies sourced from his human investor network and companies sourced from the A.I. platform was about 75% to 25%. Yet in this fund, the firm is aiming for a 50/50 split.
They argue that using an A.I. platform can help combat some of the bias and favoritism that makes venture funding notoriously inequitable. “We can tap into the networks of high-quality first-time founders in underserved parts of Europe, and then on evaluation, we have much more control over the things that we take into account and the things we don't take into account throughout the whole process,” said Arpaia.
While founders have stressed and strategized about getting the attention of VCs, maybe in the future they can wait for A.I. to find them.
Giphy sold for a discount: Yesterday Meta agreed to sell animated images startup Giphy to Shutterstock for $53 million, a stark markdown from the $315 million Meta paid to acquire the company in 2020. The announcement comes after a yearslong legal battle with U.K. antitrust regulators after they ordered Meta to sell the company in 2021 because they argued the deal limited competition. The sale comes at a time when antitrust scrutiny is intensifying on tech companies in both Europe and the U.S.
Lucy Brewster
Twitter: @lucyrbrewster
Email: Lucille.brewster@fortune.com
Submit a deal for the Term Sheet newsletter here.
Jackson Fordyce curated the deals section of today’s newsletter.
