What is one specific practice or strategic choice that has most enabled rapid revenue growth in your AI company or portfolio, and why?

Most answers emphasize leveraging proprietary data, rapid iteration, or specialization to drive growth, though some focus on renting core models, maximizing usage, or building developer-focused products.

10 answers

Answers

  • Reid Hoffman

    Co-Founder, Executive Board Chair · Manas Ai

    Close feedback loops quickly

    Reid Hoffman says a founder's job is to constantly close as many feedback loops as possible as quickly as possible, then add leverage, describing an AI-native company as a set of feedback loops hill-climbing against an objective function.

    22nd Century Frontier® · 22nd Century Frontier® · Read
  • David Haber

    Board Member · Partnership For New York City

    Rent Core AI Models

    David Haber says companies should rent core AI models from the best provider rather than build them, because whatever they build themselves will get commoditized in six months as frontier models advance.

    Fintech Leaders · Fintech Leaders · Read
  • Ben Horowitz

    Horowitz · Andreessen Horowitz

    Build a developer stack

    Ben Horowitz cites ElevenLabs getting into the developer stack and building the better product for developers, which gave it built-in customers and a data feedback loop that improved its model.

    The answer starts at 41:47.

    a16z Podcast · Listen on Apple Podcasts
  • Michael Mignano

    Partner · Lightspeed Venture Partners

    Build products quickly and cheaply

    Michael Mignano says it is getting dramatically easier to build a product, noting his firm incubated a product built very quickly and cheaply, though he frames this as raising the bar for standing out rather than as a revenue driver.

    The answer starts at 17:07.

    Sourcery · Listen on Apple Podcasts
  • Daniel Ives

    Partner and Senior Managing Director, Tech Analyst

    Maximize token consumption

    Daniel Ives says his belief is that token consumption leads revenue, so if token consumption ramps aggressively, revenue follows.

    The answer starts at 6:55.

    The Pomp Podcast · Listen on Apple Podcasts
  • Peter H. Diamandis

    Founder & Curator · Abundance360

    Proprietary Domain Data

    Peter H. Diamandis says entrepreneurs who have specific use-case data can build a company by using it to tune or train a model, because such specialist models have a good chance of outperforming foundation lab models in that use case.

    The answer starts at 14:01.

    Moonshots with Peter Diamandis · Listen on Apple Podcasts
  • Ben Gaffney

    Deputy General Counsel · Openai

    Proprietary data

    Ben Gaffney says having proprietary data is one of the things that sets a company up for success, because it is much harder to replicate than what the technology can do by itself.

    Shift AI · Shift AI · Listen at 53:18
  • Anish Acharya

    General Partner · Andreessen Horowitz

    Specialize deeply, charge high prices

    Anish Acharya says narrow startups should build small, go deep and charge a lot, arguing that specialization is a new moat because collapsing software costs let a company go so much deeper for one customer that competitors would need years of roadmap to match it.

    EO · EO · Listen at 8:19
  • Sunil Dhaliwal

    Founder and General Partner · Amplify Partners

    Invest in strong technical founders

    Dhaliwal says his firm's through line is backing technical founders with insight into emerging problems.

    N/A · N/A · Listen at 14:18
  • David Weisburd

    Co-Founder · Weisburd Pierce

    Build AI agents

    David Weisburd says that starting from scratch today he would build AI agents to handle different parts of the venture capitalist role before hiring any humans, and would make everything legible to AI and LLMs to create a learning loop for the agents.

    How I Invest with David Weisburd · How I Invest with David Weisburd · Listen at 35:32

More Generative AI questions

View all →