Listen now on YouTube, Apple Podcasts, or Spotify.
👤 Guest Bio
Tomasz Tunguz is the founder and General Partner at Theory Ventures. He launched Theory’s debut $230M fund in 2023 and has since continued to scale the firm with additional fundraising. Before starting Theory, Tomasz spent over a decade at Redpoint Ventures, where he backed companies like Looker (acquired by Google), Dremio, and StackRox.
He’s also well known for his long-running blog on SaaS metrics, which helped define how founders and investors measure growth. Today, he’s applying that same analytical rigor to AI, helping founders navigate the shift from SaaS playbooks to AI-native moats.
PRESENTED BY AUTOSKILLS
Take your team from AI-curious to AI-ready in months days
Autoskills helps teams to go from AI-curious to AI-ready with:
→ AI acceleration sprints
→ Fractional AI automation engineers to build AI workflows
→ Custom AI transformations
Teams that work with Autoskills cut hours of repetitive work, identify high-ROI use cases, and leave with the confidence (and playbook) to scale AI responsibly.
Limited to 3 clients per quarter - book a free AI Readiness Audit today!
🎙Episode Intro
For years, Tomasz Tunguz has been the go-to voice for understanding SaaS metrics. His frameworks shaped how founders pitched VCs and how investors evaluated companies. But with the rise of AI, those playbooks are being rewritten.
We recorded this conversation a few months ago, but the insights are just as relevant today. In this episode, Tomasz explains how AI changes everything from go-to-market to company valuation, why new kinds of moats are emerging, and what founders should focus on in the next 24 months. Whether you’re building or investing, this is a roadmap for navigating the AI-native era.
⏱ What’s Covered
(00:00) Tomasz’s journey from Redpoint to founding Theory Ventures
(06:45) Why AI changes the metrics that matter for early-stage companies
(12:10) Data moats, workflow moats, and what actually endures
(18:35) Why distribution is everything in an AI-first world
(25:42) Lessons from Looker, Dremio, and other portfolio companies
(32:50) The role of open-source in AI infrastructure
(40:15) How VCs are adapting their own processes with AI
(47:02) The next two years: what Tomasz is most excited (and worried) about
💡 Key Takeaways
AI metrics ≠ SaaS metrics. Traditional benchmarks (CAC, LTV, net retention) miss how fast AI adoption curves can bend.
Data is still king, but distribution is queen. Proprietary data helps, but the best products will win by getting into workflows faster.
Moats are shifting. In SaaS, switching costs were high; in AI, users will experiment widely, so habit and integration matter more.
Founders should prioritize velocity. Early AI companies that iterate faster on model use cases gain compounding advantages.
Investors must rethink diligence. Evaluating AI startups requires new lenses - evaluating workflow embedding, model costs, and adoption loops.
Exercises you can run this week
Redesign your metrics dashboard (60 min)
→ Swap out one SaaS-era metric (e.g., CAC, LTV, net retention) for an AI-native metric (e.g., inference cost per user, adoption curve velocity, or eval pass rate).
→ Ask: Does this new metric change how we’d prioritize features?Moat-mapping workshop (90 min, cross-functional team)
→ List your company’s existing moats (data, brand, switching costs, distribution).
→ Stress-test each: Will this still hold in a world where users experiment widely with AI tools?
→ Identify at least one workflow moat (where you can deeply embed AI into daily use).Distribution-first sprint (2 hrs)
→ Pick one AI feature and map three ways it could get into users’ hands faster (API, Slack integration, open-source demo, template marketplace).
→ Rank them by speed-to-ship, then commit to testing the top one this quarter.Investor diligence role-play (45 min)
→ Pretend you’re a VC evaluating your own AI product.
→ Ask: Are our workflows sticky? Do we have proprietary data that compounds? Are our model costs sustainable at scale?
→ Use gaps as a roadmap for what to fix before your next board or fundraising conversation.
📚 References & resources
Claude — Anthropic’s AI assistant; Tomasz uses it to interrogate and compare research papers, a workflow you can adopt for faster learning.
DuckDB — Lightweight, open-source analytics database you can run locally for quick queries and prototyping.
Hex — Collaborative notebook and analytics tool for data teams; helps teams explore, visualize, and share insights faster.
🔗 Where to Find Tomasz
Blog: tomasztunguz.com
X/Twitter: @ttunguz
LinkedIn: Tomasz Tunguz
👉 If you found this episode useful, share it with a teammate or founder navigating AI adoption.
And subscribe to AI Ready for weekly lessons on how leaders are making AI real at scale.
Until next time,
Haroon