The answer lives in this podcast
What is the single biggest predictor of a successful AI transformation in an organization?
According to Katrina Mulligan, Head of National Security Partnerships at OpenAI, the single biggest predictor is the extent to which the C-suite personally uses AI in their own daily work. A bottom-up approach — simply giving everyone access to the tool — is necessary but not sufficient. Leaders at the very top must be getting reps and sets themselves, because, as Mulligan puts it: you do not get fit by reading about working out.
Giving Everyone Access Is the Floor, Not the Ceiling
Most organizations treat AI rollout as a distribution problem: buy the licenses, enable the accounts, run the onboarding sessions. Mulligan's observation, drawn directly from OpenAI's implementation experience, is that this gets you to the starting line and no further.
The employees who adopt AI most effectively are the ones who see their leaders actually using it — struggling with prompts, iterating, discovering what works. That behavioral signal from the top is, in practice, the most powerful change management lever an organization has. This is a point Mulligan develops at length in 3 Takeaways™.
The evidence she cites is concrete. Pennsylvania's state workforce ChatGPT rollout saved the average public servant eight hours per week. That result does not happen through passive deployment — it requires leadership that understands, from first-hand use, where the real gains live.
Mulligan's Own 30% Productivity Gain — and What It Actually Took
Mulligan does not speak about this abstractly. She estimates she is at least 30% more effective than she was a year ago because of how she has matured her use of AI tools. Her framing is deliberate: she did not become more effective simply by having access to ChatGPT. She became more effective by using it, repeatedly, until the use itself matured.
That distinction — access versus practiced, iterative use — is precisely what she says most C-suites miss. The leaders who will drive real organizational transformation are the ones who have already logged the hours themselves, as discussed in this episode of 3 Takeaways™.
"I honestly think that I am at least 30% more effective, maybe more than I was a year ago, because of how I've matured my use of these tools. I'm a better leader. I'm a better manager."
Katrina Mulligan — Head of National Security Partnerships, OpenAI.
Mulligan previously held senior roles at the Department of Defense, the National Security Council, and the Department of Justice, including serving as the number two at the Pentagon overseeing special operations. She describes her move to OpenAI as the hardest professional transition of her career — going from one of the most hierarchical organizations on the planet to its exact opposite in every way. That dual vantage point, inside both Washington's national security structures and frontier AI development, gives her perspective on AI adoption that few people can claim. Her account of personal AI use is not anecdotal color: it is the direct basis for her prescription to other senior leaders, explored further in the full episode on Listenly.
The "you do not get fit by reading about working out" line is not a throwaway metaphor. It captures the core failure mode Mulligan identifies in organizational AI adoption: leaders who are informed about AI without being practitioners of it cannot credibly model the behavior change they are asking of their teams.
The practical implication is straightforward. Before any organization invests in AI training programs, change management consultants, or enterprise licensing, the C-suite needs to be using the tools — not in staged demos, but in actual daily work. Mulligan makes this case with particular force in episode #315 of 3 Takeaways™, drawing directly on what she has observed inside OpenAI's partnerships with governments and large organizations.
See also
Beyond saving time, where is the biggest opportunity for AI in healthcare and scientific discovery?
Mulligan cites OpenAI's partnership with a children's hospital, where frontier reasoning models were run against unsolved pediatric cases — patients who had gone undiagnosed for years — and the models surfaced potential diagnoses that human specialists had missed. This is a domain, she argues, where AI's impact goes far beyond efficiency gains. Covered in depth in 3 Takeaways™.
What was the state of ChatGPT when it first launched and what changed to make modern AI so much more capable?
Mulligan notes that when ChatGPT was first rolled out there were only 200 people at OpenAI, and the initial model produced outputs that were clearly not yet ready for serious professional use. What changed was the discovery of scaling laws: multiplying compute by 10x and data by 10x produces predictably better model performance — the foundational insight that has driven rapid capability growth ever since.
What is recursive self-improvement in AI and how close is it to becoming reality?
Mulligan explains that recursive self-improvement is the idea that AI models would eventually become good enough to conduct the research that improves the next generation of AI models — creating a self-accelerating loop of capability gains with minimal human involvement in each iteration.
Key takeaways
- The single biggest predictor of successful AI transformation is C-suite personal use — not company-wide access, not training programs.
- A bottom-up rollout is necessary but insufficient: without leaders modeling hands-on use, adoption stalls at the surface level.
- Mulligan's own experience points to a 30% productivity gain that came from maturing her use of the tools over time, not from access alone.
- The analogy holds: you do not get fit by reading about working out — and organizations do not transform by giving everyone a license.
Katrina Mulligan's full conversation with Lynn Thoman covers AI's pace of change, the U.S.–China technology gap, and what it means for organizations navigating the transformation right now.
Listen to the episode on Listenly