Every Artifact Is Somebody's Promotion Packet

Here is the claim, and it is the one I am least comfortable making, which usually means it is the one worth writing.

Most of the documents a product manager produces are not for the customer, and they are not really for the team either. They are for the promotion. The PRD, the strategy deck, the roadmap, the six-pager, the readout: these are the artifacts a calibration committee points at when they decide what level you are. The document is not the work. The document is the evidence. And AI is about to collapse the evidence.

The short version

The PM career ladder runs on artifacts, but not for the reason people assume. Documents are how promotions get justified. When you get leveled, someone in a calibration room points at the PRD you authored or the strategy you wrote and argues it demonstrates the scope of your next level. The artifact is the currency of advancement. That is why AI collapsing the document is not a productivity story, it is a career-structure story. When a spec that used to take three weeks takes an afternoon, its existence stops being proof of work, because everyone can now see it was never the work. The people most exposed are managers whose leverage was grading artifacts and senior ICs whose reputation was being the best writer in the org. The fix is to rebuild leveling around outcomes and judgment, which is harder to see and harder to game, which is why most orgs will avoid it and keep promoting the best document producers into a role where documents no longer exist.

The uncomfortable observation

Think about how a PM actually gets promoted. It is almost never "she made the number go up," because attribution is contested and the number moved for ten reasons. It is "she drove the strategy for X," and the proof of driving is a document with her name on it. The artifact is legible. Impact is arguable. Calibration committees run on the legible thing.

So the incentive, quietly, for years, has been to produce impressive artifacts. Not fraudulently. Most PMs believe the doc is the job, because the system taught them that it was. The twelve-page spec felt like real work because it took real time and produced a real object you could hold up in a review and say "I made this." The effort was genuine. The object was genuine. But the object's main job, the one nobody says out loud, was to exist in a form a promotion committee could point at.

I have sat in the calibration rooms. I have been the one pointing. "Look at the depth of this PRD, that is a Staff-level artifact." I said versions of that sentence for years and meant them. The document was our unit of evidence because it was the most legible thing we had. That is not a scandal. It is just how performance systems work: they reward what they can see, and a document is easy to see.

What AI actually removes

The comforting version of the AI story is that it takes the busywork so PMs can do strategy. Half true, and the half that is true is the part everyone repeats because it is pleasant. The half nobody wants to say: a lot of the "busywork" was the evidence. When the artifact collapses from three weeks to an afternoon, you do not just save two weeks and three days. You remove the object the ladder was calibrated to measure.

This is the mechanism that makes the Product Builder shift actually disruptive to careers, not just to workflows. The unit of work moved from the document to the working prototype and the eval. The prototype is a better artifact in every way that matters to a customer. But it is a worse artifact for the promotion system, because it is small, it is fast, and it does not carry the visible weight the committee learned to read as seniority. A prototype that took an afternoon and settled a three-week debate does not look like Staff work. It looks like you got lucky. The evidence got better and less legible at the same time, and legibility is what the ladder was built on.

When the document took three weeks, its heft was mistaken for proof of judgment. When it takes an afternoon, the heft is gone and only the judgment is left, and judgment was always the hard thing to see. AI did not create the gap between legible output and real impact. It removed the cover that hid it.

Who this actually threatens

The individual PM is not the most exposed person here, which surprises people. Two other roles are repriced harder.

The first is the manager whose leverage was grading artifacts. If your value as a director was that you gave the best feedback on a PRD, and the PRD is now a prototype-and-eval workflow you do not personally understand, your leverage just evaporated. You cannot calibrate work you cannot read. A lot of middle management in product exists to review documents, and the document is going away. That is a bigger career story than any individual contributor's, and it is why the director layer is under pressure first.

The second is the senior IC whose reputation was being the best writer in the building. That person spent a decade becoming excellent at the exact artifact that just lost its status. Their skill is real. It is also now abundant, because the thing they were uniquely good at producing, a clean and thorough document, is the thing a model produces for free. The reputation was collateralized against a currency that is being devalued.

I am not saying either of these people is now useless. I am saying they are mispriced, and mispricing gets corrected, and it is better to see it coming than to be surprised by it in a reorg.

What the new evidence base looks like

If the artifact is no longer the currency, something has to replace it, and the honest answer is that the replacement is harder to fake. That is a feature, and it is also why orgs will drag their feet.

The new evidence is a shipped surface with a metric attached, and your fingerprints on why it moved. It is an eval rubric you own that caught a real failure before it reached a customer. It is a bet you made under uncertainty, with the reasoning written down before the outcome was known, so anyone can check whether you were right for the right reason or just lucky. It is a decision log, not a document library.

Notice what all of those have in common: they are records of judgment under uncertainty, not records of effort. The old ladder measured how much you produced. The new one has to measure whether you were right, and how you knew, and that is a much harder thing to grade in a calibration room. It cannot be skimmed in the ten minutes before the meeting. It requires a manager who understands the work well enough to tell a good bet from a lucky one. Most leveling systems are not built for that, and rebuilding them is the real work of this transition, more than any tool adoption. I wrote the level-by-level version of what that evidence looks like in what replaces the PM ladder.

Try this week

Two things, one for you and one about your org.

For you: open your last performance review or promo packet and count how much of the evidence is a document you authored versus an outcome you caused. If it is mostly documents, you are collateralized in the currency that is being devalued, and you have time to change that. Start building the other kind of evidence now, a shipped thing with a number, a decision with its reasoning attached.

About your org: ask your manager one direct question. "When the deliverable is a prototype and an eval instead of a spec, what gets me to the next level?" If they have a real answer, you are somewhere that has thought about this. If they do not, that is not your failure to solve alone. It is the org telling you its ladder still runs on a currency the work has already left behind, and that is exactly the thing you want to know before the next calibration cycle, not after.