Skip to content
EO StudioEO Studio

35M Users. $100M ARR. My 10-Year Bet Was Right. | Otter.ai, Sam Liang

"Shakespeare never left a voice note. That's the problem Sam Liang has spent 10 years solving." Sam Liang, Co-founder and CEO of Otter.ai, started in 2016 when recording a meeting felt invasive and sharing notes felt strange. He had a PhD from Stanford, and a belief that most people thought was wrong: voice would become the primary interface for business intelligence. He built his own speech recognition from scratch instead of using third-party APIs. He watched competitors come and go. Today, Otter.ai has 35 million users and $100M ARR. Here's what he got right about voice AI before anyone else did and why he thinks 95% of the world is still just getting started. 00:00 Intro 01:24 The Bet Nobody Believed In 03:54 Why He Refused to Use Third-Party APIs 06:01 The Next Interface Isn't a Screen 🔗 Read the full transcription of Sam’s interview: https://www.eomag.io/article/otter-ai-sam-liang?utm_source=youtube&utm_medium=description EO stands for Entrepreneur& Opportunities. As we're looking to feature more inspiring stories of entrepreneurs all over the world, don't hesitate to contact us at partner@eoeoeo.net X | @eostudi0 LinkedIn | @EO STUDIO Newsletter | https://www.eomag.io/subscribe?utm_source=youtube&utm_medium=description

Sam Liangguest
Mar 24, 20268mWatch on YouTube ↗

CHAPTERS

  1. 0:00 – 0:30

    AI note-taking in education: capturing “lost” human voice knowledge

    Sam opens by challenging schools that ban AI tools like Otter, arguing the education system is outdated. He frames voice as humanity’s most natural interface and laments how much spoken knowledge has historically been lost because it wasn’t captured.

    • Professors restricting AI tools reflects “old thinking” about education
    • Modern learning should allow students to use AI to learn more effectively
    • Most spoken knowledge throughout history was never recorded or searchable
    • Voice is a uniquely rich medium for preserving human intelligence
  2. 0:30 – 1:01

    Otter’s evolution and scale: from transcription to agentic enterprise knowledge

    He introduces Otter.ai and describes its progression from a transcription app to an AI meeting assistant, and now toward a meeting-centric enterprise knowledge base with workflows. He anchors the story with traction: tens of millions of users and $100M+ ARR.

    • Otter started as transcription and expanded into an AI meeting assistant
    • Current direction: enterprise knowledge base built around meeting content
    • Adding agentic workflows on top of captured meeting intelligence
    • Business traction: 35M+ users and over $100M ARR
    • Enterprises adopting Otter to manage and reuse meeting content
  3. 1:01 – 1:31

    Learning to think big: Stanford PhD mentorship and the “world-changing” lens

    Sam credits his Stanford PhD advisor, David Sherrington, with shaping his ability to identify world-scale impact opportunities. He points to the advisor’s early belief in Larry Page and Sergey Brin as a model for bold conviction.

    • Stanford PhD experience influenced his long-term ambition
    • Mentor emphasized identifying what can change the world
    • Example: early support for Page and Brin before Google was proven
    • Takeaway: build with a big, durable vision
  4. 1:31 – 2:01

    Google Maps platform years: foundational product and infrastructure experience

    He recounts his time at Google (2006–2010) leading the Google Maps location platform. The role built his understanding of large-scale platforms and product infrastructure.

    • Worked at Google from 2006 to 2010
    • Led Google Maps location platform efforts
    • Gained experience building scalable platforms
    • Prepared him for founding deeper-tech startups
  5. 2:01 – 2:32

    First startup: persistent mobile sensing, personalization, and acquisition

    After leaving Google, Sam founded a mobile startup focused on location tracking and persistent sensing to understand user behavior. The company was successfully acquired, setting the stage for a larger next bet.

    • Quit Google in 2010 to start a mobile company in Palo Alto
    • Built early location tracking plus persistent mobile sensing
    • Used behavioral signals to personalize mobile services
    • Outcome: startup was successfully acquired
  6. 2:32 – 3:02

    The bet nobody believed in: record everything and share meeting knowledge

    Sam explains the core frustration that led to Otter: meetings generated critical knowledge that was hard to remember and harder to share. In 2016, the idea of recording meetings and broadly sharing notes made many people uncomfortable, but he believed culture would shift.

    • Meetings created valuable information that was easily forgotten
    • Sharing meeting knowledge across teams was difficult with personal note-taking
    • 2016 vision: record meetings and make knowledge shareable
    • Anticipated cultural change despite discomfort around recording
    • Built product to enable and normalize the new workflow
  7. 3:02 – 3:32

    Adoption dynamics: winning early users to pull the market forward

    He outlines how new product adoption follows a curve: you can’t convince everyone at once. Early adopters get compounding benefits and become internal champions who drive broader adoption.

    • Not everyone can be convinced—adoption follows a predictable curve
    • Early adopters gain productivity and effectiveness advantages
    • Champions demonstrate value and help persuade colleagues
    • Big opportunities often start before the majority believes
  8. 3:32 – 4:03

    Why deep tech matters: differentiation beyond “easy” meeting note apps

    Sam argues that to build a generational company, a startup needs deep technology roots. He notes that today many can build basic note takers quickly, so defensibility comes from foundational AI and engineering depth.

    • Generational companies are built on deep technology (e.g., Google)
    • Basic meeting note tools can be quickly replicated
    • Differentiation is the central startup challenge
    • Deep engineering enables new categories, not just features
  9. 4:03 – 5:03

    Refusing third-party APIs: owning speech recognition for speed, control, and cost

    He explains why Otter chose to build its own speech recognition instead of waiting for or paying for external APIs. Despite resource constraints and competitive risk, owning the core tech enabled lower costs and more strategic control.

    • In 2016, waiting for external APIs would have made them late
    • Building in-house speech recognition carried major risk and uncertainty
    • Competing with Google/Microsoft required conviction and focus
    • Owning core tech lowers unit cost and supports a robust free tier
    • Control of the stack improves long-term differentiation
  10. 5:03 – 6:04

    Hard unsolved problems: modeling real multi-speaker conversation at scale

    Sam highlights that meeting intelligence isn’t just transcription; it requires understanding multi-party dialogue and interaction. He argues these problems demand internal AI research and can’t be fully solved by generic third-party services.

    • Key challenge: modeling human conversation, not just converting speech to text
    • Meetings involve overlapping speakers and complex interaction patterns
    • Leveraging hundreds of millions of voice data requires specialized modeling
    • Deep AI scientists and custom research are necessary
    • Third-party APIs can’t address core, differentiated problems end-to-end
  11. 6:04 – 7:04

    The next interface isn’t a screen: voice as the primary enterprise input

    He predicts technology will shift behavior again—similar to email giving way to Slack—toward voice-first work. In his view, people will write and type less, speaking naturally while AI handles the documentation and synthesis.

    • New technologies reshape communication behavior over time
    • Voice maturity will push voice to become a primary enterprise interface
    • Typing and keyboard use will decline as speech becomes easier than writing
    • AI already drafts documents, emails, and posts—trend will accelerate
    • Otter’s thesis: voice becomes the interface to business intelligence
  12. 7:04 – 8:09

    Building for the next decade: huge remaining market and founder endurance

    Sam emphasizes that most of the world still hasn’t adopted tools like Otter, so the opportunity is long-term. He closes with a founder mindset on persistence—comparing startup difficulty to marathons and stressing endurance through expected challenges.

    • Still massive headroom: 90–99% of the world hasn’t adopted Otter-like tools
    • Focus on the next 10 years, not short-term cycles
    • Generational companies require long time horizons
    • Startups are harder than marathons—persistence is essential
    • Personal endurance practices (marathons) help manage stress and sustain effort

Get more out of YouTube videos.

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