Skip to content

HubSpot’s Co-Founder on Finding Your Next Startup Idea

a16z Podcast

58m
Transcript

Original podcast

Where to listen

Summary

Written by a model from the transcript; not yet reviewed.

Joe Schmidt and HubSpot co-founder Dharma Shah discuss software development, startup ideas, and how artificial intelligence changes business building. Dharma explains how HubSpot evolved from a marketing tool into an SMB platform by focusing on narrow use cases and solving entire customer problems rather than isolated features. They examine where startups can build durable value outside the paths of frontier AI labs by focusing on proprietary workflows, customer context, and domain expertise. Drawing on HubSpot's history and modern experiments like Us Spot, Dharma argues that high-level abstractions and faster iteration make this the best era for builders.

  • Founders should start with a narrow aperture and focus intensely on solving a specific initial problem rather than stressing over a massive total addressable market. 8:30
  • Entrepreneurs should avoid building features where frontier labs can easily absorb them or where the product concept only improves with the word yet. 24:40
  • Founders must prioritize structured deterministic database queries over expensive AI inference whenever exact data retrieval is needed. 35:33
  • Established companies can foster true innovation by creating isolated internal startup labs that operate with minimal bureaucratic baggage. 42:50

Chapters

  1. 0:00
    Podcast Intro and Host Greeting

    The episode opens with the host introducing the podcast and setting the stage for the discussion.

  2. 1:23
    HubSpot’s Evolution and AI Vision

    Dharma discusses HubSpot’s journey from marketing tools to a CRM platform and its plans for AI integration.

  3. 2:24
    Founders Should Start Narrow and Iterate Fast

    The conversation turns to startup strategy, emphasizing quick learning loops and focusing on small markets before scaling.

  4. 3:36
    HubSpot’s Product Journey: From Legal to CRM

    Dharma explains how HubSpot pivoted from a legal ERP idea to a marketing suite, then to a full CRM platform based on customer needs.

  5. 5:29
    AI as Infrastructure vs. Competition

    Dharma outlines a framework for building around AI, treating it as infrastructure and avoiding direct competition with frontier labs.

  6. 6:24
    Building Ecosystems and Developer Friendliness

    The hosts discuss how HubSpot has cultivated a developer ecosystem and the importance of openness and trust.

  7. 41:04
    Exploring Alternative Tech Choices

    The speaker reflects on how different tech decisions could have altered the product’s trajectory and discusses testing those alternatives.

  8. 42:50
    HubSpot Next Innovation Lab

    The conversation turns to HubSpot’s internal startup program, its structure, and its role in fostering experimentation.

  9. 43:35
    Us Spot: One-Person CRM Experiment

    Details about the Us Spot project, its target market, rapid launch, and early monetization strategy are shared.

  10. 45:41
    Scaling or Absorbing New Ideas

    Discussion on how successful experiments can be scaled or integrated into the main product line at HubSpot.

  11. 49:45
    Market Opportunity for One-Person Companies

    The speaker highlights census data showing a surge in single-person businesses and the potential for tailored software solutions.

  12. 52:15
    Vertical SaaS and AI Potential

    Exploration of how AI expands the scope of vertical software and the opportunities it creates for small customers.

Highlights

  • 2:24

    Best Times and Fog of Warinsight

    Dharma describes running a public software company through a massive technological transformation as both the best of times and a period filled with the fog of war.

  • 4:06

    Increasing Levels of Software Abstractioninsight

    Tracing software development history from machine language to modern AI, Dharma explains how each jump increases abstraction and empowers more people to build.

  • 6:47

    Rejecting Spreadsheet Idea Selectioninsight

    Dharma argues that rigorous analytical spreadsheets cannot predict startup success, and the only real way to test an idea is to ship it.

  • 15:45

    Discovering the CRM Opportunitydata

    HubSpot discovered that 60 percent of their paying marketing customers were not using any CRM because existing tools were too intimidating and expensive.

  • 23:31

    Avoiding the Frontier Model Pathstrategy

    Dharma advises founders to stay off the well-paved road where frontier labs compete, focusing instead on structural constraints or areas with incentive constraints.

  • 43:35

    Building Us Spot as an Experimentstory

    Dharma shares details on Us Spot, a brand-new 30-day-old one-person CRM designed as a startup within HubSpot to test monetization and second-brain concepts.

  • 50:33

    Tracking US Company Formationsdata

    Dharma cites US Census data showing record company formations in recent years, with over 70 percent predicted to remain one-person companies.

Quotes

  • “The interface is a means to an end.”
    Unnamed speaker · 0:13
  • “We had to do it all in one offering.”
    Unnamed speaker · 0:30
  • “One of the things I think the common mistakes entrepreneurs make in terms of kind of idea selection is that they think it's a matter of rigorous analysis, that we're going to have this kind of 18-column spreadsheet that's going to give me all the coefficients that out it's going to spit the best possible idea.”
    Unnamed speaker · 6:41
  • “If you can just append the word yet to it and it still holds water, that's a problem.”
    Unnamed speaker · 24:40
  • “Anything that can be solved with a structured query deterministically, solved with a structured query deterministically. Don't use inference to solve something that a query can solve.”
    Unnamed speaker · 35:29
  • “And we sort of carried that idea through now.”
    Unnamed speaker · 43:23
  • “The answer is yes. It does work.”
    Unnamed speaker · 44:34
  • “I want to learn like the pricing lessons and some of the things that we want to learn more about kind of monetization pricing”
    Unnamed speaker · 50:00
  • “In 2025, we had the most number of new company formation that the United States has ever seen.”
    Unnamed speaker · 50:33
  • “There's never been a better time to build a software company.”
    Unnamed speaker · 55:03

About this episode

As frontier AI labs expand into applications, where should startups build? a16z Partner Joe Schmidt joins HubSpot co-founder and CTO Dharmesh Shah to explore why the software opportunity is growing, and how founders can stay off the “Yellow Brick Road” the labs are racing down. Drawing on HubSpot’s evolution, Dharmesh explains why founders should start with a narrow market, learn quickly, and solve the whole customer problem. They discuss where application companies can build lasting value as models become infrastructure: workflows, customer context, and the systems where work happens. They also explore HubSpot’s AI experiments, the rise of one-person businesses, and why cheaper software creation could unlock markets that previously couldn’t support a software company.

Transcript

0:00
58:57

12,935 words · 3 speakers

Topics

  • HubSpot
  • AI
  • CRM
  • ChatGPT
  • a16z
  • Us Spot
  • OpenAI
  • HubSpot Next
  • tokens
  • Twitter

Page updated · transcript and summary are machine-generated