LLenny's Podcast
← All frameworks
LeadershipCaitlin Kalinowski (ex–OpenAI, Meta, Apple)

Zero-to-One Team Building

For brand-new categories, hire generalists, adjacent-domain experts, and AI-native new grads.

Difficulty
Moderate
Time to result
~months to results
Steps
4
Confidence
90%

When you're building something that has never existed, you can't staff it with people who've done exactly this before — because no one has. Kalinowski's hiring recipe for zero-to-one work blends strong generalists who transfer skills across fields, a mix of build-the-new and scale-the-known experience, adjacent domains (like autonomous vehicles for robotics), and genuinely AI-native young engineers who can teach the rest of the team. Mission alignment unifies the mix, and gut-feel for 'spark' is the final filter.

Origin

Caitlin Kalinowski, on hiring for OpenAI's robotics program and prior 0-to-1 hardware teams.

Core principles

  • 01For a truly new category, exact prior experience doesn't exist — stop requiring it
  • 02Generalists who adapt lessons across fields outperform narrow specialists in new domains
  • 03You need both zero-to-one builders and people who've scaled other things
  • 04Only genuinely AI-native engineers (often ~20-21) can teach a team AI-first workflows
  • 05Mission alignment unifies teams drawn from very different worlds

How to run it

  1. 1

    Recruit strong generalists who transfer across fields

    Prioritize hybrid people who can adapt what they learned in one field to a new one. In new categories the 'exact same thing' simply doesn't exist, so adaptability beats a narrow track record.

    Pro tip Mine adjacent domains: for robotics, autonomous-vehicle talent brings the sensing stack, safety trade-offs, and hardcore engineering.

  2. 2

    Balance zero-to-one builders with proven scalers

    Hire some people with experience building the new thing and some with experience scaling other things. You need both capabilities on the same team, even if the team is smaller than teams used to be.

  3. 3

    Bring in genuinely AI-native new grads to teach the team

    Only engineers who grew up using AI so natively it's baked into their process — often around 20-21 — can show senior people how to work AI-first. They approach problems completely differently and move faster. Hire them not just to execute but to teach.

    Pro tip Pair AI-native juniors with senior engineers deliberately — the goal is knowledge transfer in both directions, and you must have both seniority levels.

    Watch out Assuming AI erases junior roles is wrong; without junior + senior mix you stop building new technologists and lose the AI-native way of thinking.

  4. 4

    Filter on mission alignment, then gut-feel spark

    Require alignment to the mission — it unifies people coming from AI-research and hardware worlds who otherwise miscommunicate. Once every hard criterion is checked, use gut feel for the 'spark': genuine motivation, hunger to learn, willingness to update their view, and a drive to win.

In the wild

Staffing OpenAI's robotics program from autonomous vehicles

Building a robotics program for a robot that could move through the world at the scale of millions — something no one had done — Kalinowski looked to autonomous-vehicle talent for the sensing stack and safety trade-offs, combined hardcore roboticists, and added AI-native young engineers.

She reports attracting some of the top robotics talent in the world to a from-scratch program.

Common mistakes

Requiring exact prior experience for a novel category

Insisting on people who've built 'this exact thing' is impossible when the thing has never existed, and screens out the generalists who can actually adapt to it.

Assuming AI has erased the need for junior hires

You must have both senior and junior engineers; without junior AI-native talent you can't learn AI-first workflows or build the next generation of technologists.

Is it for you?

Best for

Founders and hiring managers staffing a from-scratch team in an emerging, undefined technical category

Not ideal for

Filling a well-defined role in a mature domain where deep, exact specialist experience is directly available and preferable

From the transcript

some people who have experience building the thing that you're building that's new

1:20:00

the only AI native people essentially who use AI so natively that it's like baked into their engineering process are 20 years old

1:20:00

I rely a lot, Lenny, on my gut feel for people, assuming everything else that I'm looking for has been checked.

1:23:00

From the episode

Why we’re at the beginning of the AI hardware boom

Caitlin Kalinowski (ex–OpenAI, Meta, Apple)