Using AI to Determine The Right Investment
Artificial intelligence is no longer a thing of the future. It is becoming a necessity in many industries as a driving force for growth. In the VC world, a number of investors are already using AI to determine whether or not a potential investment is worth exploring.
In our most recent DiffuseTap event, we examined this growing phenomenon with our speakers, Miguel Gonzalez, co-founder of AI-enabled company and investor discovery startup FounderNest, and Brad Zapp, co-founder of VC firm Connetic Ventures. DiffuseTap is a weekly virtual event hosted by Diffuse that is part networking (you’ll meet at least a half dozen high calibre startup players) and part purposeful (you’ll DiffuseTap new ideas).
During our session, Brad and Miguel talked to us about how they are (1) utilizing AI to ensure that they are making the right investments and (2) connecting investors with companies who share the same vision.
Using AI to assess value in people
According to Brad, when Connetic started incorporating AI into the company screening process, it was out of necessity. With early stage being the focus of Connetic, the challenge was how to accurately assess a startup’s potential, given they usually lack sufficient data for analysis.
Brad and his team decided that they needed to take a systematic approach to evaluating founders. However, being based in Kentucky, where there are not too many venture deals to fund, meeting founders “only through planes, trains, and automobiles” would not yield a large enough assessment pool. So instead, they sought a viable solution to help with their screening process.
But how do you dig into the DNA of a human being? How can you determine whether a person is “hardwired” to succeed?
“Entrepreneurship is really hard, and if you just don’t have the inherent behavioral DNA inside of you, it’s difficult,” Brad explained. They soon realized that there wasn’t an assessment tool in the marketplace robust enough to answer these questions.
Thus, Wendal was born. Wendal is a proprietary behavioral survey tool that places participants in a “vector analysis”, to help Connetic answer questions like: Is this person an “enterpriser” that can really lead an organization? Is this person tilted toward sales or product advocacy? Is this a team member, a collaborator? Is this a specialist that’s really formal and likes details?
FounderNest, on the other hand, uses AI to discover companies and investors and to determine if there is a fit between the two that would warrant a “double opt-in” introduction.
Similar to Connetic’s Wendal bot, FounderNest uses an AI technology to analyze the key strengths of a team. As Miguel explained: “We’ll look at signals such as the academic and professional background of the founders. Have they worked together in the past and for how long? Do they have any relevant experience in the industry in which their current company is operating? Have they started a company in the past, scaled it, and eventually exited?”
FounderNest then supplements that team-driven analysis with traction metrics, such as the company’s revenue, CAC, LTV, and other industry-specific metrics. This combination of founder background and track record evaluation shapes a defensible picture for the VC connections to be made.
So how accurate is AI?
Using AI to grade investments is one thing. But obviously, investors also want to know if these AI-driven assessments are accurate. According to Brad, Connetic has an alpha-beta error to estimate how exact these AI models are, and they continuously improve on it.
Connetic runs different tests simultaneously. A recent test assesses funding opportunities similarly to Morningstar.com, using a star rating system, wherein one star is potentially a bad investment and five stars shows great promise.
“What I can tell you is that only 14% of our ‘one-star’ companies are doing well, compared to 74% of the ‘five-star’ companies.” Brad also qualified, “I think we’re about a year, maybe 18 months away, to get a lot more comfortable. But we’re running this weekly.”
Miguel shared that FounderNest has yet to reach that point of sophistication wherein “it makes sense for us to look at the beta or alpha of our model”. Instead, they look at conversion rates on the two sides of the marketplace — investor and investee — and based on that, determine whether their “double opt-in” connections are a right fit.
“On the founder side, 80% of the investors that we identify for our startups are accepted by the company. And then 25% of the companies that we discovered for investors are accepted by those investors. So, in essence, we have a 22% conversion rate from the investors that we suggest to our company to the number of ‘double opt-in’ connections that we made,” Miguel reported.
AI zooms in on minority-backed companies
Connetic has seen the potential of AI beyond just geographics. Leveraging Wendal, they’re looking into new opportunities within minority-owned businesses, thus, breaking down barriers and encouraging diversity in the VC market.
As Brad explained this welcome discovery, “Everybody knows the statistics about the lack of funds getting to female, minority-backed companies. Everybody knows what’s happening, at least in the U.S. today, with all the institutionalized social injustice examples being brought to the forefront. Here’s what I can tell you from a data standpoint: out of the 1,500 or so companies that have had conversations with Wendal, approximately 40% of those are minority-owned businesses.
“And I think if I’m right, that’s about on the national average. The AI bot program itself has been recommending about 43%, so it’s basically saying ‘Hey, once we carve out [your prospects], 43% of acceptable investments actually are minorities.’ And that’s super interesting because it’s deeply correlated with the funnel,” Brad shared.
From that perspective, Wendal goes past deep-seated biases that investors, as humans, can’t simply overcome. This allows an opportunity to level out the playing field and is potentially a huge step towards the right direction in the VC world.”
“You can’t erase your own biases,” Brad elaborates further. “But isn’t it nice to have a machine tell you ‘hey, here’s these four, five, six, seven other companies that maybe you wouldn’t have picked up on, but you really need to take a closer look.’ That’s been the coolest thing so far. Personally in Connetic, once we got into diligence, we ended up funding about 48 or 49% [of minority-backed companies].”
On the other hand, Miguel said FounderNest has decided not to explore that side of their matchmaking algorithm because aside from it being a big investment, it might pose some ethical questions on their end.
“It has some ethical -slash- legal issues that may be involved when asking that kind of ethical information from [the] founder. So we’ve decided to hold off on that until making sure that there is really an opportunity to make an impact in this space,” Miguel explained.
Nonetheless, FounderNest has decided to power B+ Ventures, the largest initiative in the U.S. focused on connecting Black, Hispanic, and immigrant founders with investors. “So while we are not doing this directly through our own matchmaking, we are definitely all about supporting these kinds of initiatives that are trying to bridge that gap,” Miguel said.
Limitations of AI screening
Using AI to assess human qualities poses some limitations, especially given the ambitious goals both companies are trying to achieve. As Brad noted, “How do you assess where there is a personality fit between the founder and the investor? How do you make sure that there is that personal chemistry between the two sides?”
Miguel said FounderNest has explored these questions in many different ways. “Because it’s such an important ingredient of the equation, there is not an easy answer. At the end of the day, it is so subjective, it really depends on the tone, on the body language of the founder when you meet with the founder for the first time…And also on the dynamics between the founders when you are meeting with them, and the way they communicate. That is really hard to figure out through a model that works for everybody across the board.”
Echoing Miguel’s sentiment, Brad revealed that they loved how codification was such a hard problem. In fact, they’ve taken it on with help from an industrial psychologist at Xavier University. “We pitched him our idea on how we wanted to accomplish this, [and] he bit it. And so we partnered my team’s crazy thoughts, industrial psychology, [and] software development. And then that’s how we developed our startup DNA app.”
Brad expounded further, “So, you know, when you think about the founders of WeWork, Uber, and Theranos, I bet they were awesome. And I bet if we were in that room [when they pitched their companies], we all would want to invest. And those might have been huge mistakes, so we want to tackle this problem. It is really hard. We have put math behind it. Do we have enough data to say we’ve nailed it? No. But we’re doing it.”
More funds are using AI for screening
AI tools at maturity create a great opportunity for both investors and startups to increase their probability of success. And that’s why we’re seeing several funds today using AI for screening. However, we have yet to see AI fully incorporated in the process across the board. COVID may just have been the push that the industry needed.
“I do see it as a trend.” Brad noted. “Early on, pre-COVID, when we came on the scene in early 2019, and we unleashed the Wendal bot, we probably had 15 per cent serious pushback. People were like, ‘what do you mean I can’t have a coffee with you? Talk with you, pitch you all this stuff?’ Because they had been trained by their mentors or whatever on how to go through that traditional VC process.
“But then COVID hit, and all of a sudden everyone had to be digital. And all of a sudden, we were sitting here with like, 18 months under our belt, where we had been in digital VC for a long, long time. And all of a sudden the responses were pretty surprising. Like in April, we had a record high. I’ve never seen anything like it. It was 100% uptick for us from a deal flow.”
Miguel, meanwhile, has observed that in the early stage space, there are still very few VCs building AI-based models for valuation. “At the end of the day, earlier stage evaluations of the companies are dictated by the market itself. So, how much demand you’re able to create around your company, and who is behind that demand. [Do] those VCs or angels talk to your angels? If they [do], most likely your valuation is going to go up.”
Brad agreed with Miguel, pointing out that AI-based assessments can only go so far; however, they do have great potential, and we’re all ready for it, he remarked. “Do I think in the next two or three years everyone will just let Wendal, the bot, pick investments for them? Absolutely not. I think that would be a mistake. But do I think automation, machine learning, data, and tech platforms will help enable all VCs? And even startups? Yes, I do.”
Brad Zapp is the co-founder and partner at Connetic Ventures, anchored in Cincinnati, Ohio. Connetic is currently valued at $30M and focuses on investing in early stage, innovation-driven startups in North America and Europe.
Miguel Gonzalez is the co-founder and COO/CPO at FounderNest, a Silicon Valley company that helps founders identify the right investors for their companies and investors discover hidden gems that fit with their investment thesis. Currently, FounderNest has made around 1,500 connections and has nearly 1,600 VCs from all over the world under its belt, 70 percent of which are top tier VC firms, mostly in Silicon Valley.
Diffuse incubates and runs emerging VC funds. If you have an investment thesis that you would like to launch into a fund, get in touch today.Find an event near you