What practice keeps your AI business durable, not hype-driven?

Zach Lloyd

Founder / CEO · Warp

Listen to the market

Zach Lloyd says the principle is to listen to the market, a lesson from spending too much time on a preconceived notion of what would make the business work; he warns not to get anchored on biases.

The answer starts at 52:01.

Cory Janssen

Co-Founder · Altaml

Focus on industry expertise

Cory Janssen says he focuses on specific industries rather than the hype around foundational models, partnering with companies that have industry knowledge to build an edge on data so new entrants find it hard to compete.

The answer starts at 25:26.

Sam Stephenson

Founder · Granola

Focus on daily user value

Sam Stephenson says his principle is to hold himself accountable not just to shipping but to whether the person actually gets daily value out of it, and to not stop until that goal is achieved rather than just the shipping goal.

Sam Altman

Chief Executive Officer · OpenAI

Empowering builders

Sam Altman says empowering builders, rather than trying to be the one giant evil mega corp, is important, citing OpenAI's commitment to give developers the same tools it uses.

Azeem Azhar

Founder · Exponential View

Real sales revenue

Azeem Azhar says the revenue number is disproportionately important, because real sales from real customers will get almost any business out of whatever trouble it's in.

The answer starts at 58:50.

Anshu Sharma

Co-founder & CEO · Skyflow

70% hiring, 30% firing

Anshu Sharma says his practical step is hiring, putting it at 70% hiring and 30% firing, because everything else takes care of itself.

Jeff Finkelstein

Owner · Customer Paradigm

Focus on end‑user needs

Jeff Finkelstein says he focuses on the end user's needs, because AI software fails when people treat it as a shiny object and ignore what matters to the end customer.

Euwyn Poon

Founder · Orbital - Space Data Centers

Only build things that feel difficult

Euwyn Poon says his principle is to build only things that feel difficult, because company building should feel incredibly difficult and if it feels easy it likely isn't the right thing.

Greg Brockman

President, Chairman, & Co-Founder · Openai

Stay close to technical details

Greg Brockman says to stay close to the technical details, which he calls the best way to ensure safety guarantees, arguing that rules around architectures miss the boat.

The answer starts at 23:04.

Dan Martell

Founder & CEO · Martell Ventures

Red teaming

Dan Martell recommends red teaming: attacking your idea, asking why it won't work and where you're kidding yourself, because it helps validate ideas and avoid building something nobody wants.

The answer starts at 18:31.

Ben Cera

Founder · Polsia

Focus on storytelling

Ben Cera says he focuses on storytelling, citing his weekly YouTube channel where he tells the story of building his company, which he says helps him gain attention more cheaply than running ads.

The answer starts at 6:16.

Eric Simons

CEO & Founder · Bolt.New

Play long-term games with long-term people

Eric Simons says his principle is to play long-term games with long-term people, saying short-term strategies are high-risk for him and anyone working for him.

Grant Lee

Co-Founder/CEO · Gamma

Building end‑to‑end workflow

Grant Lee says the practice is building an end-to-end workflow for a business, not just winning the first mile. He argues that going deep into the workflow and staying dedicated to the space can produce a purpose-built solution over time.

Andrew Ng

Founder · Deeplearning.Ai

Talk to customers

Andrew Ng says people who can talk to customers and get a sense of what to build are positioned to build with AI, because the cost of building has plummeted and deciding what to build is now the bottleneck.

Tom Bilyeu

CEO · Impact Theory

Pre‑established metric

Tom Bilyeu says to have a pre-established metric you'll respond to, decided when emotionally sober, so emotions don't drive your decisions.

John Brewton

Writer & Founder · Operating by John Brewton

Build what couldn't be built before

John Brewton says the useful question is not what model launched this week but what can we build now that we couldn't build before, since when models improve or cross thresholds, new things become buildable.

Anastasios Angelopoulos

Chief Executive Officer · Arena

Entrenchment within enterprise

Anastasios Angelopoulos says it's all about entrenchment within enterprise, adding that if so, it's golden, and if not, to be more nervous.

Alex Levin

Co-Founder & CEO · Regal

AI should always be identified as AI

Alex Levin says one of the three principles he proposes is that AI should always be identified as AI, whether through upfront disclosure, verbal cues, or invisible stamps.

Cal Al-Dhubaib

Head of AI and Data Science · Further

Trust Engineering

Cal Al-Dhubaib points to "Trust Engineering," the subject of his LinkedIn Learning course, saying AI controls differ from traditional security and that governing these tools is a continuous process, not a one-time exercise.

Qasar Younis

Founder, CEO · Applied Intuition

Make money

Qasar Younis says the best way for an AI company to survive is to make money, so you don't need to raise money and will still be there when markets grow.