For most of your career, the way to become more valuable was to row harder and better than the people around you. Ship more. Ship cleaner. Be the one whose work came back with the fewest corrections. Execution quality was the ladder, and climbing it was mostly a matter of effort and craft. Call that rowing.
Tara Seshan, who leads product for Codex and ChatGPT Work at OpenAI, puts the shift as a move from rowing to steering. As AI takes on more of the execution, the human contribution becomes judgement and direction: deciding where the boat goes, how fast, and when the current heading is wrong. Her teams build for where the models will be in two or three months rather than where they are today, on the assumption that execution capacity keeps getting cheaper. What does not get cheaper is knowing what to do with it.
Ambition as a constraint
Seshan’s most intriguing claim is that ambition becomes the bottleneck for a company, and that raising other people’s ambition is now part of the job.
The claim is worth taking literally. When execution is scarce, a team’s ceiling is set by how much it can build. When execution is abundant, the ceiling is set by how good its answer is to the question what is worth building. A team that ships quickly but aims low does not benefit much from faster shipping. It just arrives at a mediocre outcome sooner, with more polish on it.
I have seen this already. The capacity is there to attempt something genuinely hard, and the plan that comes back is last year’s roadmap with the timelines compressed. A team that was going to ship four features now proposes to ship seven, none of which changes the shape of the product or the position of the business. The tooling improved but the ambition did not move yet. That gap is now the thing that decides how much value an organisation gets from AI, and most leadership teams are not managing it as a variable at all.
Ambition is also uncomfortable to raise, which is part of why it stays low. Asking a team to aim higher invites the response that the current plan is already full. It is: of execution that used to be expensive and no longer is. Someone has to look at a full plan and say that most of it is now the easy part.
Steering is a different skill
Steering is not rowing done at a higher level. It is a distinct capability: choosing a direction under real ambiguity, holding the bar for what counts as good, and recognising early when a course needs to be abandoned rather than pushed harder. None of these three is trivial, or even well understood, in most teams.
Being excellent at execution does not build that muscle. The engineer who was the best by throughput, the product manager who shipped the most, the manager whose team never missed: their reputation was earned rowing. Some of them are also strong at steering, but if they are, that’s a bonus. Steering is a completely separate thing, and the transition exposes who has been quietly relying on execution quality to stand in for judgement.
Steering also resists the two moves that made execution manageable. You cannot delegate it downward the way you hand off a task. You can delegate a piece of the work, but the moment you delegate the choice of direction, you have either kept the real decision or handed over your job. And you cannot fake it with activity. A full calendar is evidence that you are busy, not that the last big call you made was the right one, and the gap between those two things is where a lot of senior careers have quietly been living.
The tell that you are rowing when you should be steering is simple: you can describe what your team did last quarter in detail, and you cannot clearly say why that was the right thing to have done rather than something else.
What to practise
If steering is the scarce skill, treat it like one and train it deliberately.
- Make the direction calls explicit. Every cycle contains decisions about what not to do, which are usually made by default and never written down. Write them down, with the reasoning. That is the record you learn from.
- Raise the bar on done. If a prototype now takes an afternoon instead of a sprint, “it works” is no longer an achievement. The standard should move to whether it is worth shipping at all.
- Spend the freed capacity on harder problems, not more of the same. The temptation is to produce more of what you already produce. The opportunity is to attempt the thing that was previously out of reach.
None of this is new advice for people who were already good at leading. What is new is that it stopped being optional. The rowing is being handled.
The boat is faster than it has ever been. The only question left is whether you know where you are pointing it.
Leave a Reply