Views: 269
AI is not only changing the way startups are built, but also how VCs operate. Here at Remagine Ventures, we are actively navigating this shift as the venture capital industry, which took over 70 years to begin moving away from a manual “handcraft” approach, rapidly embraces data-driven models where algorithms and automation augment human investors.
Let’s dive into how automation tools, AI, and data science are revolutionizing the venture capital investment process, fund management, and portfolio support, with concrete examples from the funds leading the charge.
The Investment Process: Sourcing and Diligence on Autopilot
Sourcing and evaluating startups used to be highly reliant on serendipitous networking. Today, VCs are building bespoke tech stacks to automate deal flow and run deep diligence at scale.
For instance, Hustle Fund is able to review a staggering 700 deals every month by heavily relying on automation tools like Zapier, Typeform, Process Street, Airtable, and Pipedrive to ensure no steps in their process are missed. Meanwhile, Presidio Ventures has utilised GPT-4 to help develop machine learning sourcing models, and EQT uses OpenAI’s ADA embedding model to extract keywords and interact with data more efficiently.
We are also seeing the rise of agentic AI workflows (outside source material). While many developers are experimenting with autonomous coding tools like Claude code (outside source material), VCs are utilizing similar autonomous agents for research. Untapped Capital, for example, utilizes BabyAGI to run efficient, automated research that searches multiple web sources and internal CRMs to combine findings into a single report. Untapped also pairs Wokelo.ai with their own proprietary AI analyst to perform sentiment analysis on Product Hunt reviews, extract tech stacks via BuiltWith, and apply basic reasoning via OpenAI to understand market risks.
Taking it to the extreme, Koble.ai operates as a purist “Quant VC” firm. Disrupting the traditional model, they aim to entirely remove the human from the investment decision equation, relying purely on quantitative data and algorithms to select startups. Similarly, Rule30 has built a probabilistic decision engine that analyzes founders based on their career trajectories and education, ranking them into the top 10%, 3%, and 0.1% percentiles. To avoid human bias, Rule30 operates with no human in their Investment Committee. On the other side of the spectrum, Boardy has introduced an AI “superconnector” that holds direct voice conversations with founders and formally recommends investments to a human committee.
Fund Management: The Rise of the Data-Driven LP
The management of venture capital itself, including how Limited Partners (LPs) select which VC funds to back, is also getting a major technological upgrade.
Level VC, a modern fund of funds, heavily differentiates its management strategy by building internal technology to ingest massive amounts of VC transaction and public market data. They use this data to construct a massive knowledge graph that maps out the venture capital ecosystem, treating each investor as a node. By running algorithms on this graph, Level VC can analyze the strength of an investor’s relationships and score managers to predict future performance, effectively turning LP investing from a game of intuition into a data-backed science. They even pass this advantage on to the emerging managers they back by providing them with custom data applications for daily operations.
Supercharging Portfolio Support
Many VCs claim to add value post-investment, but tracking metrics and supporting dozens of portfolio companies manually is incredibly taxing. Automation and AI are helping VCs deliver on their promises at scale.
Hustle Fund leverages ChatGPT to parse through unstructured founder updates, automatically extracting crucial data like revenue, runway, and growth. This parsed data is then funneled via integrations into Slack summaries and Airtable to build out internal benchmarking for the portfolio.
Davidovs Venture Collective (DVC) takes unstructured data automation a step further. When a founder sends an email update or a PDF pitch deck, DVC uses a tool built by one of their LPs called “AInalyst”. It uses LayoutLMv3 to recognize text and charts, and then feeds it into GPT-4 via Langchain to extract necessary information and automatically generate a “MiniMemo” for the team.
Additionally, DVC feeds routine founder updates into GPT-4, using prompt engineering to automatically identify a founder’s needs and convert them into actionable tasks. The AI then suggests which community members are best suited to help based on database tags. To support founders during fundraising, DVC also utilizes Perplexity.ai to seamlessly generate highly effective, personalised double opt-in introduction emails.
VC is being rewired by AI
While venture capital remains a relationship-driven business at its core, the industry is rapidly transitioning. By connecting robust CRMs with Claude Code or OpenAI’s Codex and deploying advanced AI models, VCs can uncover better opportunities, process massive amounts of information, and offer unprecedented support to their founders.
He is a former General Partner at Google Ventures (GV) in Europe, former head of Google for Entrepreneurs in Europe, and founding head of Campus London, Google’s first startup hub. Eze writes on Israeli tech, venture capital, artificial intelligence, and founder strategy.
He is also the founder of Techbikers, a nonprofit that brings together the startup ecosystem on cycling challenges in support of Room to Read.