A client landed a verbal offer from OpenAI
The work that made it possible wasn't about AI fluency
A client received a verbal offer from OpenAI!
A recruiter reached out to them while we were deep in the prep for a different role at a different lab. They took the call, used the same answers they had been rehearsing, and received a verbal offer about 2 weeks later.
We have been working together for about a year. It started when we worked on negotiations for their current role, and then we worked on their onboarding to fast-track their credibility and influence in the new role. After that we stayed together and kept working to up-level how they think and how they communicate. So when this recruiter call came in the middle of their interview prep, the prep itself did not feel like a heavy lift. It felt like a continuation of the work we had been doing.
They were already a model PM before we started. Years of shipping at the frontier of AI. Fluent in model behavior, evals, and agentic systems. The technical content was never the question.
What we worked on was not what they knew. It was how they talked about what they knew.
The turning point was when I asked them a warm-up question and asked them to tell me about something they had recently shipped. They walked me through it in order. They told me what the product was, when it launched, who used it, and what the metrics looked like.
It was accurate and thorough, but something was still missing.
I told them it sounded like a Wikipedia entry.
They laughed and said, “This is exactly what I do at work every day. I don’t talk to my CEO like this. I don’t talk to my skip like this. I come in with what I’m thinking about, what I’m betting on, what I’d do differently. Why does my brain switch into Wikipedia mode the second someone says ‘tell me about’?”
That is the gap I work on with leaders. It is not about what they know, and it is not about whether they can think strategically. It is about what happens when the conversation gets framed as an interview.
At work, the conversation is forward-looking by default.
Their skip walks into a one-on-one already knowing the context and the history, and the implicit question is always some version of “what are you betting on now, what are you thinking about, what would you change.” This client answers that question every week without thinking about it. They walk in and say, “Here’s what’s been bothering me about this product area. Here’s the bet I want to make. Here’s what I’d cut to fund it.” Their skip sees a senior leader every single week.
An interview asks a different question. It is backward-looking by default.
The interviewer does not know the context or the history, and the implicit question is always some version of “describe what you have done so I can decide whether you are good enough to do it again here.” That framing pulls people toward narration. They start listing what they shipped, when it launched, what the metrics looked like, and how it went.
The same person who walks into a one-on-one with their skip and names a bet, walks into an interview and starts narrating a project.
That is the identity shift.
Same brain, projects, resume. They did not lose their strategic thinking the moment they sat down for an interview. They answered the question they were asked, and the question an interview asks is a description question. So they described.
What we did over the next several sessions was rebuild various stories so the version of them that shows up at work was the version walking into the interview. Here is what that looked like in practice.
The Wikipedia version of a story sounds like this:
“I led the launch of our agentic workflows product. We shipped in Q3. Adoption hit 40% of eligible users in the first quarter and we saw a 22% lift in retention. I worked closely with researchers, engineering, and design to get it out the door. I also did some of the coding myself and pushed a few PRs to production.”
That is descriptive language. It tells the interviewer what happened. It is accurate, and it would pass at L5 and maybe L6 (Meta and Google leveling, see levels.fyi for translations). At L7 and above, it sounds like someone executing a plan rather than someone making one.
The strategic version of the same story sounds like this:
“The hardest call we had to make was whether to build agentic workflows or invest in better single-turn quality. Most of our usage was still single-turn, and the easy bet was to double down there. My read was that we were measuring demand for what users already knew how to ask for, not demand for what they would ask for once the product could do more. The bet I made was that agentic was the wedge into a different kind of user behavior, and that single-turn usage was a ceiling we were close to maxing out, not a floor. If I had read it wrong, we would have spent months building something nobody used. The 22% retention lift came from users we acquired who would not have stayed for single-turn, which is the population I was building for.”
That is strategic language.
It opens with the call you had to make.
It articulates what most of the data was telling you and why you read it differently from the people around you.
It articulates the bet you placed, what would have made you wrong, and what made the outcome land where it did.
The metrics are still in the story, but they are evidence for your judgment instead of the point of the story.
Employers do hire for experience, but experience is the minimum requirement. Experience is the low bar that gets you the interview.
The high bar is whether:
you can make the right call when the data is ambiguous
your judgment is repeatable
the way you think travels into a new environment that is different and likely more ambiguous than the one you came from.
The strategic version of a story makes all of that visible to the interviewer.
The Wikipedia version hides it underneath the description of the work and leaves the interviewer guessing whether the judgment underneath was yours or someone else’s.
By the time this client had the OpenAI call, this muscle was already built. They had not prepped specifically for OpenAI, and they had not needed to. After working on how they think and how they talk about their thinking, they could take the interview and sounded like the senior leader they are at work, because that is who they are now in interviews too.
The content was already there. They had years of frontier AI work behind them. What changed was learning to walk into an interview as the person they already are at work, instead of as the polished resume version of that person.
And when the right opportunity comes, the application to work together is here.
For L7-L9 PM interviews: https://nancychu.hbportal.co/public/acing_interviews
For L7-L9 interviews across other roles: https://nancychu.hbportal.co/public/67c378d9a7c8e4001fea1a9b/1-acing_interviews
If you have been reading for a while and you have never commented, I would love to know: Where does your brain switch into Wikipedia mode? Drop a comment below. I read every one :)
Nancy