Skip to content
The Twenty Minute VCThe Twenty Minute VC

Jeff Seibert: Why OpenAI Will Become an Infrastructure Play | E1085

Every single 20VC episode is recorded with Riverside.FM. It is the one product that I could not live without. Try it today here (https://creators.riverside.fm/20VC) and use the code 20VC for 15% off. ----------------------------------------------- Jeff Seibert is the Founder & CEO @ Digits, building the future of AI-powered accounting. Digits have raised funding from the likes of Peter Fenton @ Benchmark and 20VC. Jeff previously served as Twitter’s Head of Consumer Product, a position he came to following the acquisition of his prior company, Crashlytics. Today, Crashlytics is the de-facto mobile crash reporting solution for iOS and Android and runs on over 6 Billion monthly active smartphones worldwide. ----------------------------------------------- Timestamps: (0:00) Intro (00:30) Investor Meetings and Twitter's Impact (01:14) Childhood Ambitions and Product Design (04:03) Entrepreneurial Struggles and Management (08:26) CEO Accountability and Company Culture (12:50) Digits: Origin and Evolution (18:36) Business Pivots and Investor Interaction (26:17) AI Impact and OpenAI's Challenges (36:13) Data Privacy and AI Adoption Speed (41:14) Angel Investing and Startup Outcomes (54:25) Quick-Fire Round ----------------------------------------------- In Today’s Episode with Jeff Seibert We Discuss: 1. The Art of the Pivot: What are Jeff’s biggest pieces of advice to founders pivoting? How do you know when you have enough data to make the decision to pivot? What are the single biggest mistakes founders make when pivoting? 2. AI: Who Wins and Who Loses: Why does Jeff believe that OpenAI will transition into an infrastructure play? What are the most significant challenges OpenAI will face moving forward? Why does Jeff believe that Apple are best positioned to win in an AI world? Why does Jeff believe that Google are the most vulnerable incumbent? What would Jeff do if he was CEO of Google? 3. LLMs: What Happens Now: Will we see the commoditization of LLMs? What are the biggest misconceptions people have on training and fine-tuning LLMs? Will we see LLMs increasingly specialise to vertical-specific models or will they remain horizontal? What is the difference between a thick and a thin wrapper when building on top of LLMs? 4. Angel Portfolio in Review: How many angel checks has Jeff written? How many failed? How many home runs? Does Jeff believe that company valuations are being kept artificially high? How did Jeff make 200x selling through the secondary market for a now failing company? What are Jeff’s three biggest pieces of advice for angels today? ----------------------------------------------- Subscribe on Spotify: https://open.spotify.com/show/3j2KMcZTtgTNBKwtZBMHvl?si=85bc9196860e4466 Subscribe on Apple Podcasts: https://podcasts.apple.com/us/podcast/the-twenty-minute-vc-20vc-venture-capital-startup/id958230465 Follow Harry Stebbings on Twitter: https://twitter.com/HarryStebbings Follow Jeff Seibert on Twitter: https://twitter.com/jeffseibert Follow 20VC on Instagram: https://www.instagram.com/20vchq Follow 20VC on TikTok: https://www.tiktok.com/@20vc_tok Visit our Website: https://www.20vc.com Subscribe to our Newsletter: https://www.thetwentyminutevc.com/contact ----------------------------------------------- #VentureCapital #JeffSeibert #Digits #HarryStebbings

Jeff SeibertguestHarry Stebbingshost
Nov 21, 20231h 1mWatch on YouTube ↗

At a glance

WHAT IT’S REALLY ABOUT

OpenAI, Infrastructure, and Execution: Jeff Seibert on AI and Startups

  1. Jeff Seibert, serial founder and CEO of Digits, explains how large language models will commoditize and why OpenAI is likely to become an infrastructure provider similar to AWS rather than a full-stack app company. He argues that startups must treat AI as a tool, not a product, and focus on deep customer problems, strong execution, and decisive pivots rather than thin wrappers on foundation models.
  2. Seibert shares hard-won lessons from building Crashlytics, operating at Twitter, and founding Digits: speed, conviction, intentional execution, and culture-driven feedback loops matter far more than most founders realize. He also details how Digits navigated a multi-year R&D wall, a pivot into tooling, and then a second pivot back to its original vision once GPT-class models made real-time accounting finally possible.
  3. The conversation expands into the future of AI platforms, with Seibert predicting commoditized LLMs, an "Android to OpenAI" open-source ecosystem, and Apple emerging as a major on-device AI player due to control of silicon and privacy. He also assesses the vulnerability of incumbents like Google, the likely trajectory of AI adoption in enterprises, and why climate change and job displacement narratives are misunderstood.
  4. On angel investing, Seibert candidly shares portfolio data, the dominance of outliers, and why disciplined, uniform check sizes and founder grit matter more than early traction or hype. He underscores that most early-stage bets are far harder and longer-dated than investors expect, and that secondary liquidity often beats waiting for IPO outcomes.

IDEAS WORTH REMEMBERING

5 ideas

Treat AI as infrastructure, not the product itself.

Seibert stresses that LLMs will soon resemble databases or cloud compute—powerful but commoditized tools. Startups should win by solving hard, enduring customer problems and building workflow depth, not by merely scripting calls to GPT.

Avoid building thin wrappers on top of OpenAI or any single model.

If your core value is just orchestrating prompts into a general LLM and you built it in a few weeks, you are likely on OpenAI’s roadmap and will be “Sherlocked” when they ship a native feature or the model’s base capabilities improve.

Pivots must be decisive, all-in, and backed by runway.

Seibert argues a pivot is not a side experiment; it’s a full strategic reset that requires strong founder conviction, team buy-in, and at least ~12 months of cash so the new direction can actually prove itself.

Execution quality and intentional decision-making separate enduring founders.

Most founders and managers are poor operators: they’re unintentional with time, priorities, hiring, and what they say no to. Seibert emphasizes fast, traceable decisions (e.g., a ‘24-hour hypothesis’ rule) so teams keep moving and can learn from outcomes.

Design organizations where great ICs don’t have to become managers.

Promoting top individual contributors into management by default often fails. Seibert advocates trial management periods, destigmatizing returns to IC roles, and compensation ladders that reward world-class ICs without forcing a managerial track.

WORDS WORTH SAVING

5 quotes

I view OpenAI probably evolving more into an infrastructure company like AWS.

Jeff Seibert

Startups are gonna get killed because they're very thin wrappers. If your primary product value is scripting GPT, that's a thin wrapper.

Jeff Seibert

Most founders aren't intentional about how they operate the business… If you don't have conviction, you're really gonna struggle as a founder.

Jeff Seibert

What kills companies is uncertainty… I would rather see founders take one to-the-moon bet on one new direction.

Jeff Seibert

Google's by far the most vulnerable… It is way more effective to kill your own golden goose than watch someone else do it.

Jeff Seibert

Why OpenAI will likely evolve into an infrastructure provider (AWS-analog)Commoditization and specialization of LLMs and the rise of open-source modelsApple’s potential advantage in on-device AI via custom silicon and privacyStartup execution: speed, intentionality, pivots, and culture of feedbackBuilding Digits: from R&D wall to pivots enabled by GPT-era modelsFounders, management, and organizational design (IC vs manager tracks, accountability)Angel investing dynamics, portfolio realities, and lessons from outlier outcomes

High quality AI-generated summary created from speaker-labeled transcript.

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.

Add to Chrome