jobs don't have a fitting room - by Hilary Gridley
SubscribeSign in
jobs don't have a fitting room<br>prototype everything, including your career
Hilary Gridley<br>Jul 15, 2026
Share
Quick note: Due to popular demand, I re-opened enrollment for the July cohort of How to be a Supermanager with AI until midnight tonight . Enroll now for lifetime access to the course content.<br>I recently wrote about a simple way to figure out what you want. Unfortunately, knowing what you want is often only one part of the challenge. Once you know what you want, you have to set out to find what you want.<br>Finding what you want is hard because the beautiful theories inside our heads do not always match up with real life. You think, I would love a more creative job, and then you get a more creative job, and you find out that “creative” corporate environments are home to some pretty terrible managers, and actually that makes your life much worse.<br>Henrik Karlsson writes and speaks about making contact with reality: “A painter does not want technique to get between them and the canvas. The faster you can collide your ideas against reality, the faster you get feedback. By increasing the speed at which you can act on the context, trying new things will become cheaper for you—and so you will take more risks, and extract more information from the context.”<br>I love this sentiment. But what happens when even reality moves too slowly? If you are hoping to make better choices about, say, your career, you may find yourself in a painfully slow feedback loop: get a job, wait until you’ve settled in, weather a few big projects, get a promotion, wait for the new role to settle in....suddenly two years have gone by before you find yourself in a place to evaluate what, exactly, suits you or doesn’t suit you about the role.<br>You can, of course, collect more data by talking to people about their jobs and trying to imagine what you would like or not like based on their reports.<br>But ideally, you would be able to “try on” jobs much more rapidly than reality will often allow. I have not yet found an equivalent of a fitting room for jobs. But I do have the next best thing: prototypes.
Subscribe
One of the bigger changes in my life since I started using AI is that it has wildly expanded the scope of decisions that I try to solve with “prototype thinking.”<br>In product development, we use prototypes to help us make better decisions about what to build because there is a very large gap between a good-seeming idea and the fully realized execution of that idea. Instead of describing an idea to someone, we can give them a concrete, if low-fidelity, example of the idea to observe how they interact with it. This starts to close the gap between theory and reality.<br>Importantly, the prototype is not the thing. It is simply trying to communicate some key attributes of the thing so that you can make more informed decisions in the process of making the thing real.<br>I am constantly asking myself what I can prototype. I prototype essays: I might want to explore ten different possible structures for an essay to follow; in my head I think one structure will be best, but of course it’s hard to know without having something to react to. So I ask the AI to play out a few different introductions; these immediately go straight into the trash, but seeing concrete examples has moved my own thinking forward.<br>I prototype living in different neighborhoods.<br>I have recently spoken to a number of people who are looking for new jobs, and are feeling unsure of what direction to take their career. To them I say:<br>Prototype jobs!
Here is an example of a prototype for an open AI Product Director role at Abridge:<br>Tuesday, 7:40am: the chief of cardiology at your biggest customer emails — “the new notes bury the ejection fraction.” Your dashboard disagrees: quality up three points. You spot-check six transcripts yourself before standup — the doctor’s right, the rubric rewards completeness and ignores where cardiologists actually look — then hand the fix to your evals PM and spend the day on the part only you can do: telling the ML lead you’re holding Thursday’s release, and trading him a decision-by date so it doesn’t become a war; pre-briefing Friday’s physician council so the meeting ratifies your plan instead of relitigating it; and rewriting Q3’s roadmap doc, because if the rubric missed this in cardiology, it’s missing it in orthopedics and psych too, and “audit every specialty rubric” just displaced someone’s favorite feature. The uncomfortable question you end the day with: why did a customer find this before your team did?<br>Whether it is exactly correct is beside the point. The map is not the territory. What’s important is that it is concrete and specific enough that if you compare this to a prototype for a different job, you can choose between them. This is pairwise comparison -- a staple of user research -- and its power is that it forces a choice: there’s no “it depends,” no fence to...