This article originally appeared in Forbes.
We partnered with Dr. Andre Retterath on his "Data-Driven VC Landscape 2023" report and found that while 84% of venture capital firms want to increase their efforts and resources to stand up internal data-driven initiatives, only 1% have done so today.
Meanwhile, Gartner, Inc. predicted in 2021 that "by 2025, more than 75% of VC and early-stage investor executive reviews will be informed using artificial intelligence (AI) and data analytics."
These two data points tell a compelling story about how the outlook on VC strategies—from sourcing to diligence to research—is transforming. There's a lot of excitement in private capital right now about AI, not just as an investment area but also as a force that will change the way investors do their jobs. I would argue that AI is just the latest step in an underappreciated but significant data-driven journey that the venture industry has been going on for some time.
Our survey of hundreds of investment professionals at the end of 2022 revealed that three-quarters of VCs use at least four sources of data when researching a deal. This number has risen sharply over the last few years as the technologies that automate data scraping and integration to derive insights have become more sophisticated.
Today, AI is rapidly changing the general relationship between humans and the tools they use. As data-driven investing becomes increasingly the norm, my conversations with our many VC customers worldwide have surfaced three main areas to watch and experiment in:
Automating repetitive tasks to drive productivity gains
AI is already helping to streamline the manual tasks that many VCs perform. These tasks, repetitive in nature, traditionally took up a lot of time and mental bandwidth—outreach emails, intro emails, rejection emails, market maps, industry reports, thought leadership and blog posts.
VCs are looking into AI-assisted research and writing tools powered by models like GPT-4 and Bard as low-hanging fruit, enabling them to improve productivity and allocate more time to truly value-adding work like building relationships with new founders and supporting their portfolio companies.
Retterath has advocated for VCs to use these tools to increase productivity, noting that ChatGPT's launch "changed the game" due to "the introduction of a simple and intuitive UI that in turn reduced the entry barriers for non-technical users." He said ChatGPT had improved his productivity by "at least 10x" in the area of deal sourcing and screening.
Enhanced data entry automation
Firms with in-house engineering teams are using AI models to take automated data extraction and entry to the next level—particularly around deal management workflows and sourcing. For example, a GPT-4 powered crawler can automatically categorize startups by investment status or stage, summarize meeting notes and transcripts, and more.
Currently, the barrier to entry for many firms interested in this use case is nontrivial. It requires in-house expertise to build tools that can analyze raw data and push it with the right shape and accuracy into CRMs and other platforms. However, as with everything else in this space right now, I expect that to change fast.
Peregrine Badger, the instigator at Fifty Years, uses AI to help his firm identify promising startup ideas based on Ph.D. research, explaining that people on his team "could read papers all day, every day, and just crack a tiny fraction." Since the team was manually scanning Twitter for papers, he built a tool that sources and ranks papers posted on Twitter. The highest-ranked papers are then put through a ChatGPT API in order to narrow the scope based on some fundamental questions.
Investment insights that augment human intuition
This third category is the most nascent but may also be the highest leverage to VCs with the resources and willingness to become more data-driven.
Think about the upstream parts of generating deal flow: sourcing, diligence and research. Most data-driven firms aren't just buying datasets and manually searching through them for opportunities. They're writing and training AI models to look at the attributes of their most successful investments and feeding that input back into the sourcing engines to better inform the signals they should be looking at—signals that they may not have even been aware of.
This innovative sourcing strategy can help to surface opportunities that even the best investors overlook simply due to the sheer volume of data that AI can process and uncover opportunities from compared to humans. It's still far from being the norm, but it's my prediction for where our industry is headed.
None of these use cases will fully replace human judgment, but AI is quickly becoming table stakes. Retterath provides a useful analogy for understanding how the industry is evolving:
"Data-driven approaches are like an e-bike that suddenly allows the average Joe to compete with athletes. While it’s a game changer for the broad majority of people, it has little impact for athletes as they perform naturally on this level through many years of disciplined training. Still, it might help them get up this one steep hill faster than their athlete competitors."
For VCs intent on getting up that hill faster—finding those startups the moment they become relevant to the firm's investment thesis—now is the time to invest in AI.