Screen new founders and run outreach for venture investing
Lessons from a year interviewing 40+ investors across venture, growth, hedge funds, and crossover firms. Task: Lessons from a year interviewing 40+ investors across venture, growth, hedge funds, and crossover firms I've removed all investor names and identifying details to protect the privacy of the firms and individuals I interviewed. A classmate who recently raised a Seed round told me that more than 95% of the questions she got from investors were the same. Not similar. The same. Her theory was that every investor had run the same information through Claude and gotten back the same questions. AI saved them time but erased their differentiation. The investors she chose went a level deeper, asking questions that showed an independent point of view. That tension, AI as efficiency tool versus AI as investment edge, became the focus of a year-long research project I conducted at Stanford GSB. After three years at a venture firm and a summer at a research-driven hedge fund, I wanted to understand how AI could unlock new research capabilities inside venture capital firms. I interviewed more than 40 investors across early-stage venture, growth equity, hedge funds, and crossover firms, including some building “AI native” firms, around three questions: 1. What does research mean inside an investment firm, specifically venture? 2. How are investors using AI internally? 3. Most importantly, is AI making them better investors? By research I don’t mean gathering information. I mean asking questions whose answers change the decision, and forming an evidenced opinion while knowing what remains unknown. My biggest takeaway: venture capital is adopting the playbook that transformed public-market investing over the past two decades. AI is making deep research cheaper and faster, letting small teams process far more information and raising AUM per head, and at many firms it’s shaping not just how investors work but how they decide. But when everyone has the same models, research quality, knowledge, and deal flow become table stakes. New firms have won over the past decade by bringing public-equity research into private markets Many of the most successful venture and growth firms founded in the last ten years have roots in public equities: Forerunner’s Kirsten Green came from equity research, Bond’s Mary Meeker from Morgan Stanley, Greenoaks’ Neil Mehta from D.E. Shaw, and Addition’s Lee Fixel from Tiger Global. Others from that vintage built research teams that resemble hedge funds. The growth speaks for itself: Thrive, which hired researchers early, grew from a $40M first institutional fund in 2011 to a $10B Fund X in 2026. This matters because older venture firms relied on networks and brand for deal access. Research-driven firms compete differently. One investor told me, “We aim to be the most informed investor about every company we meet and win allocation based on that.” Another firm employs former investigative journalists who interview customers, ex-employees, and industry experts, layered with data-science work, all before meeting the founder. The approach is especially powerful at Series B and later. Not everyone buys this. An investor at a more traditional venture firm told me research is not where alpha is driven: everyone knows the big categories; selecting and winning deals is the unlock. But I think that is missing how strong research can indicate which deals to select. Others have grown suspicious of expert networks, where $500-a-call incentives produce fabricated experts; one called them a lagging source of alpha because entrepreneurs are ahead of customers. Fair. But having contacts with people inside companies that aren’t through expert networks can be a real source of unique insights. There is also a cautionary tale. Tiger Global brought public-market speed and volume into private markets, and its 2021 vintage showed what happens when you import the process without the price discipline. The playbook that has worked is knowing more than everyone else, not moving faster. Why fundamental-research driven investing historically worked less well at seed Several investors I met insisted the research model breaks down at the earliest stages, though I believe AI is changing that, for two reasons: 1. There’s little to analyze. A seed-stage startup may be two people, a deck, and an idea. You can research the founders, market, and competition, but there are few business fundamentals to study, and the founders will likely pivot anyway.; tools: custom-built or unnamed agent. Human role: The original post does not say how a person stayed involved.; limitations: Prerequisites include access to company filing data and a defined investment thesis. The source warns that if everyone uses the same models, research quality becomes a baseline, not an edge. Human judgment is critical for 'taste' and relationship building, which AI cannot replicate. Integration uncertainty exists regarding how to encode proprietary 'taste' into prompts effectively without converging on standard answers..
Tools
custom-built or unnamed agent
Published on Agent Work Atlas:
Overview
Lessons from a year interviewing 40+ investors across venture, growth, hedge funds, and crossover firms
Workflow
Lessons from a year interviewing 40+ investors across venture, growth, hedge funds, and crossover firms I've removed all investor names and identifying details to protect the privacy of the firms and individuals I interviewed. A classmate who recently raised a Seed round told me that more than 95% of the questions she got from investors were the same. Not similar. The same. Her theory was that every investor had run the same information through Claude and gotten back the same questions. AI saved them time but erased their differentiation. The investors she chose went a level deeper, asking questions that showed an independent point of view. That tension, AI as efficiency tool versus AI as investment edge, became the focus of a year-long research project I conducted at Stanford GSB. After three years at a venture firm and a summer at a research-driven hedge fund, I wanted to understand how AI could unlock new research capabilities inside venture capital firms. I interviewed more than 40 investors across early-stage venture, growth equity, hedge funds, and crossover firms, including some building “AI native” firms, around three questions: 1. What does research mean inside an investment firm, specifically venture? 2. How are investors using AI internally? 3. Most importantly, is AI making them better investors? By research I don’t mean gathering information. I mean asking questions whose answers change the decision, and forming an evidenced opinion while knowing what remains unknown. My biggest takeaway: venture capital is adopting the playbook that transformed public-market investing over the past two decades. AI is making deep research cheaper and faster, letting small teams process far more information and raising AUM per head, and at many firms it’s shaping not just how investors work but how they decide. But when everyone has the same models, research quality, knowledge, and deal flow become table stakes. New firms have won over the past decade by bringing public-equity research into private markets Many of the most successful venture and growth firms founded in the last ten years have roots in public equities: Forerunner’s Kirsten Green came from equity research, Bond’s Mary Meeker from Morgan Stanley, Greenoaks’ Neil Mehta from D.E. Shaw, and Addition’s Lee Fixel from Tiger Global. Others from that vintage built research teams that resemble hedge funds. The growth speaks for itself: Thrive, which hired researchers early, grew from a $40M first institutional fund in 2011 to a $10B Fund X in 2026. This matters because older venture firms relied on networks and brand for deal access. Research-driven firms compete differently. One investor told me, “We aim to be the most informed investor about every company we meet and win allocation based on that.” Another firm employs former investigative journalists who interview customers, ex-employees, and industry experts, layered with data-science work, all before meeting the founder. The approach is especially powerful at Series B and later. Not everyone buys this. An investor at a more traditional venture firm told me research is not where alpha is driven: everyone knows the big categories; selecting and winning deals is the unlock. But I think that is missing how strong research can indicate which deals to select. Others have grown suspicious of expert networks, where $500-a-call incentives produce fabricated experts; one called them a lagging source of alpha because entrepreneurs are ahead of customers. Fair. But having contacts with people inside companies that aren’t through expert networks can be a real source of unique insights. There is also a cautionary tale. Tiger Global brought public-market speed and volume into private markets, and its 2021 vintage showed what happens when you import the process without the price discipline. The playbook that has worked is knowing more than everyone else, not moving faster. Why fundamental-research driven investing historically worked less well at seed Several investors I met insisted the research model breaks down at the earliest stages, though I believe AI is changing that, for two reasons: 1. There’s little to analyze. A seed-stage startup may be two people, a deck, and an idea. You can research the founders, market, and competition, but there are few business fundamentals to study, and the founders will likely pivot anyway.
Reported result
The source reports that generic AI usage leads to identical investor questions, eroding differentiation. While one investor's system automates sourcing and scheduling, the text does not provide specific metrics on conversion rates or time saved for this exact workflow. It notes that 'sourcing is getting commoditized' and 'research is becoming table stakes.' A measurable acceptance check would be tracking the percentage of unique, non-generic questions asked in meetings compared to a control group.
Human role
The original post does not say how a person stayed involved.
Limitations
Prerequisites include access to company filing data and a defined investment thesis. The source warns that if everyone uses the same models, research quality becomes a baseline, not an edge. Human judgment is critical for 'taste' and relationship building, which AI cannot replicate. Integration uncertainty exists regarding how to encode proprietary 'taste' into prompts effectively without converging on standard answers.
Try it
Act as a venture capital sourcing agent. I will provide [replaceable inputs: target sector, founder background criteria, and a sample list of 5 newly formed companies]. 1. Analyze the provided company data against my [replaceable inputs: investment thesis and scoring criteria]. 2. Generate a shortlist of the top 2 founders who match the profile. 3. For each, draft a personalized outreach email and 3 unique, deep-dive questions that demonstrate independent research, avoiding generic AI-generated queries. 4. Output the results in a JSON format with fields: 'founder_name', 'match_score', 'outreach_draft', 'unique_questions'. Before finalizing, perform a dry run on one sample founder and show me the output for approval. Do not send any emails or schedule meetings without my explicit confirmation. If data is missing for a founder, flag it as 'insufficient data' rather than guessing.
Source reports are not independently verified. Publication date is the catalog publication date, not the original article date.