Several months ago, we built an automated deal-screening software tool and deployed it on all incoming deals. The software reviewed all incoming deals against our investment thesis and auto-scheduled meetings with companies that were a match and passed on deals that were not. When we first started using the system, it seemed like a suitable solution for the high volume of deals we receive every month. One of the goals we hoped to achieve from using the tool was to eliminate bias from our screening process and provide more founders with the opportunity to speak with us.
Initially, the software worked exactly as intended. But after several months, we realized that the tool lacked context, imagination, and nuance. For example, the software automatically passed on a company that we manually reviewed and ultimately invested in. Because the startup didn’t look “right on paper” the software passed on the company because it lacked context about the size of the market opportunity. This begs the question:
Can AI really replace VCs such that we can (or should) completely rely on it to screen deals?
The Human Touch
The uniqueness of venture capital comes from the individual decisions each investor makes. These decisions stem from an individual’s experience, knowledge, and unique insight. Is it possible to make AI 99.99% accurate at mirroring investors’ decisions? Probably. But what if the unicorn(s) that VCs seek lie in the 0.01%?
One of our favorite podcasts is DOAC (Diary of a CEO) hosted by Steven Bartlett. In one of his episodes, he spoke with an AI expert about the likelihood of humans being ruled or destroyed by AI. Steven made a profound statement. He said (paraphrasing) “If I were creating technology that I later learned had even a 1% chance of destroying humanity, I would halt production immediately and figure out a way to eliminate the chance of that happening.”
We think the same goes for maintaining human intervention in the deal selection process in the private markets industry. If there’s even a 1% chance that the next billion-dollar deal could be missed because an AI agent rejected it, that risk is too great.
To play Devil’s Advocate, one could say the opposite. If there’s a 99.99% chance that your agent finds the next billion-dollar deal, why wouldn’t you use it? Well, if all VCs were to use an all-purpose LLM trained on similar edge cases (aka unicorns), then each VC would make the same decisions. And if all VCs make the same decisions, where will diversity be in the market? Independent decisions about different startups in the market drive healthy competition. Healthy competition is instrumental in driving innovation, which is the core of the startup world.
How We Use AI at Serac Ventures
Because of the high volume of companies sent to us by those in our VC network or from founders who reach out to us via direct email, we used Claude to create a tool that auto-populates the deal info into our Airtable database. The tool is simple. When we receive an email, we copy and paste it into the tool and it extracts/parses the information needed to fill out each Airtable field. What used to be a manual process that took 5-6 minutes per company, we can now complete in 30 to 60 seconds.
There are AI tools built into Airtable that we could and formerly used to auto respond to companies that were not a match (e.g., companies outside of our geographical target area or sector focus). But as we stated above, we stopped using those tools because occasionally we need to make exceptions. And most importantly, we have deep respect for founders and the challenges they encounter as they build their companies. For this reason alone, we believe founders deserve and appreciate a human response (even a short one). It may take us a tad bit longer to respond, but a personalized email is much more appreciated than a canned response from an AI bot.
Yes, deal screening can be automated. But we believe the decision to pass or go forward is optimized with a human in the loop. Besides, what’s the fun in startup investing if you eliminate interaction with founders early in their journey? You never know if a founder you kindly say “no” to now will remember you and return in the future with an amazing opportunity. In our opinion, treating people like people is the best way to maximize the chance of this happening.
Cheers – Kevin Moore (GP) & Anisa Faryat (Venture Fellow)
.png)
