Sarah Guo: On Her New $101M Fund; How AI Impacts Inequality; AI Startups vs Incumbents | E1007

Fund Launch & Strategy

• Sarah Tavel departed Greylock Capital after ten years to launch Conviction, a new early-stage fund focused exclusively on Artificial Intelligence.
• Conviction operates with a $100 million fund size, targeting seed and Series A investments.
• Investment check sizes range from $1 million to $10 million, with a strategy prioritizing capital efficiency over large upfront capitalizations.
• The fund aims to be a "matchmaker" for the AI community, specifically connecting founders with model access, GPUs, data, and design partners.
• Tavel declined a $500 million fundraising target to maintain structural discipline, arguing that constraints foster creativity for both founders and investors.

AI Investment Thesis

• The core conviction is that AI will enable 10–20 person teams to build billion-dollar companies, fundamentally altering the traditional requirement for large headcounts.
• Tavel identifies horizontal software as the ultimate destination but argues that specialization is currently necessary due to the technical complexity of the AI field.
• The fund avoids companies attempting to train massive foundational models from scratch, noting fewer than 10 instances where this strategy makes sense; instead, the focus is on applying existing models via APIs or fine-tuning.
• Key investment areas include legal tech (e.g., Harvey.ai), code generation beyond Copilot, RPA, and multimodal creative tools.
• Tavel notes that 41% of co-creation is currently AI-assisted, predicting a shift toward human planning with AI handling iterative execution and coding.

Market Dynamics & Economics

• Tavel argues that AI is labor-replacing, creating a larger opportunity in services markets (e.g., legal services) than traditional software markets.
• She challenges the concept of defensibility at the seed stage, asserting that investors should evaluate trajectory and founder capability rather than static moats.
• Regarding capital intensity, Tavel clarifies that while training large models is expensive, most AI startups can operate with low capital by leveraging external APIs.
• She warns that regulatory bodies face a "chasm" of knowledge compared to the speed of AI innovation, creating risks for uncalibrated policy interventions.
• Tavel expresses concern over wealth centralization driven by AI but maintains that the technology will ultimately drive abundance and productivity gains.

Competitive Landscape & Incumbents

• Microsoft is identified as the incumbent leader in AI adoption, crediting Satya Nadella and Kevin Scott for their strategic bets on OpenAI and leveraging the tech for search.
• Amazon and Apple are viewed as potentially lagging due to a lack of cutting-edge internal labs and the need for significant future investment.
• Tavel emphasizes that speed of execution is the startup's primary advantage against incumbents in an environment where "a decade happens in a year."
• She dismisses the "data moat" as a decisive factor, noting that entrepreneurs are increasingly creative in generating or collecting proprietary data.

Founder Evaluation & Mistakes

• Tavel adopts a founder-first approach, willing to invest in difficult markets (e.g., EdTech) if the team possesses the unique ability to navigate structural headwinds.
• She cites her biggest regret as not investing in Rippling and Benchling at early stages, acknowledging that founder capability can transform stagnant markets like life sciences SaaS.
• Tavel admits to having no deep technical expertise in AI, choosing instead to be transparent with founders and leverage her network for strategic support.
• She warns against "generic" idea generation, advocating for high-resolution customer conversations to identify specific, non-obvious problems.
• Tavel notes that reserves (capital set aside for follow-on investment) are "complete bullshit," arguing that market outcomes at the four-firm stage are unknowable and that early-stage funds should accept leaving money on the table.

Industry Structure & LP Relations

• Tavel predicts the death of the generalist seed VC, aligning with Hunter Walk's view that technical depth and niche networks are becoming prerequisites for success.
• She warns that subscale seed funds without a differentiated strategy will struggle to persist as LPs become more cautious.
• Tavel criticizes GP commitments as "bullshit" when funds use LP money to meet the requirement rather than co-investing personal capital.
• She advocates for founders to conduct investor references heavily, noting that many fail to check the interpersonal dynamics and politics within large multi-stage firms.
• Tavel identifies near-term AI abuse (e.g., malicious code generation) as a more immediate threat than long-term AGI safety, urging investment in defensive measures.

Short-Term Outlook

• Macro Economy: Tavel expects a "digestion period" for over-capitalized startups, predicting the 2023 landscape will remain "ugly" as companies struggle to achieve efficiency.
• Venture Strategy: She predicts a bifurcation where large multi-stage firms face "group think" risks, while small boutique firms with clear mandates will thrive.
• LP Landscape: Tavel calls for more independent thinking among LPs, criticizing the tendency to follow big brands rather than evaluating individual conviction.

Personal & Career Reflections

• Tavel's measure of success for Conviction includes achieving best-in-class venture multiples, building a beloved partnership, and being chosen by the most important entrepreneurs of the next generation.
• She views her podcast, No Priors, as a rigorous business requiring high output (three episodes weekly) to build organic distribution.
• Tavel dismisses the idea of a "perfect" multi-stage fund, highlighting the risks of interpersonal politics and seniority overriding in large partner groups.
• She expressed skepticism regarding the predictability of market outcomes, noting that intellectual narratives about structural advantages often fail against the reality of human agency.