This angel investor network is using AI to speed due diligence on promising startups

Local investors are using artificial intelligence to save time on the due diligence process.

Keiretsu Forum, a global angel investor network that includes sponsors and startups in the Philly region, announced the K4-MST AI FastTrak Due Diligence Engine earlier this month. The tool aims to save time on the first draft report so that investors can spend more time actually deciding if a company is a good fit for investment, Howard Lubert, area president of Keiretsu Forum’s mid-Atlantic division, told Technical.ly. 

“Philadelphia’s startup ecosystem thrives on strong investor-founder relationships, but due diligence has always been a bottleneck,” Lubert said. “Philly-based startups [will] get faster feedback, investors can engage with more deals, and promising companies secure capital sooner.”  

Typically, due diligence reports, which help determine how risky an investment would be, take 80 to 120 hours to complete, according to Lubert. He expects AI to cut that time in half by reading through hundreds of files, summarizing them and annotating them in reference to the original files.

The tool was trained using existing due diligence reports from a variety of sectors. Keiretsu ran over two dozen trials over the last few months with local investors and startups, like medical device company Actuated Medical, based in Bellefonte.

Plus, there are guardrails in place to make sure the information in the report is accurate to what the company provided. 

“The AI engine that we’re using does not take it upon itself to go out to the world wide web and look for information about this company,” Lubert said. “The only data that is used in creating this draft due diligence report comes from the files that the companies provide in the deal room.” 

AI gives investors more time to carefully decide their next move

Despite curiosity for AI, Lubert said there wasn’t a lot out there that fit Keiretsu’s needs. Many of the existing tools worked better for mergers and acquisitions. These tools don’t work for due diligence because they can’t accurately estimate the valuation of early-stage companies, he said. 

Eventually, it found AI company Standd.io, which specifically develops platforms for due diligence processes. Companies upload all their files onto an online deal platform. Once that information is condensed into one file, Lubert can upload it into the AI platform. It typically takes the AI engine 10 to 20 minutes to pull that information into a draft due diligence report. 

The draft also comes back annotated with links back to the original documents. So if team members have questions or want to fact-check parts of the report, they can access it.

The hope is that this tool will leave more time for discussion and making careful decisions about investments, Lubert said. 

“What we hope will transpire from that is a greater interest from our investor members to get on due diligence teams, because they won’t have to make that significant [time] commitment,” Lubert said.