The seats companies are quietly re-opening, after AI did not replace the team

A pattern has been running through hiring plans for a couple of years now. A company pauses a role, or several, on the reasonable expectation that new tooling will absorb the work. The budget is reallocated, the requisition is closed, and the remaining team is asked to cover the gap with better software. Some months later the seat quietly re-opens, sometimes under a different title, usually with a different specification.

This is not a story about the tooling failing. In most of these organizations the tooling worked, and the productivity gain was real. What was misjudged was the unit of work being automated. Tasks were absorbed. Roles were not, because a role is rarely a stack of tasks. It is a set of judgments about which tasks matter, made by someone accountable for the outcome.

What actually got absorbed

The work that moved fastest was the work that was already close to mechanical: first-draft documents, summarising long inputs, structuring unstructured data, routine analysis against a known template, first-pass code, first-pass research. In functions where a large share of the week went to that kind of production, capacity genuinely expanded.

What did not move was everything requiring a decision under ambiguity with a named person accountable for it. Which of these three customers we are prepared to lose. Whether this candidate will survive the first ninety days here specifically. Whether the forecast is optimistic because the market shifted or because a regional manager is protecting a number. Those are not drafting problems. Drafting was never the bottleneck.

The result inside a lot of teams is that output per person rose while the number of decisions requiring a senior judgment stayed flat or grew, because faster production creates more things to decide about. That is the shape of the problem that re-opens a seat.

The three ways the pause tends to fail

• ✓The senior person becomes the bottleneck. Junior capacity expands, more work reaches the person who has to approve it, and the constraint moves up rather than disappearing. The re-opened role is usually a second senior seat, not the junior one that was cut.
• ✓Quality drift goes unnoticed for two or three quarters. Output that is eighty-five percent right looks fine in isolation and compounds badly. The seat that comes back is often a review or ownership function that nobody thought of as a role until it was missing.
• ✓The institutional knowledge leaves. When a team is thinned on the assumption that tooling covers it, what usually leaves is not the task capacity but the person who knew why the process was built that way. That is the most expensive of the three to replace, and the hardest to write a specification for.

Tasks were absorbed. Roles were not, because a role is rarely a stack of tasks. It is a set of judgments about which tasks matter, made by someone accountable for the outcome.

The specification has changed, and most of them have not caught up

The seats coming back are not the seats that were cut. The job description that was shelved eighteen months ago usually describes a volume of production that no longer needs a person. Reposting it produces candidates optimised for work the company no longer needs done by hand, and it filters out the people who would actually be effective.

Three shifts show up repeatedly in the specifications that work.

• ✓Fluency with the tooling is now a baseline, not a differentiator. Candidates who use these tools well are common. Screening hard on it selects for enthusiasm rather than capability, and the interesting question is not whether someone uses the tools but whether they can tell when the output is wrong.
• ✓Judgment and accountability carry a premium. The scarce profile is the person who can look at plausible-sounding output and identify the assumption that does not hold. That capability comes from having owned outcomes, which is why these searches skew more senior than the roles they replace.
• ✓Range matters more than depth in a narrow lane. Smaller teams with more capacity per person need people who can operate across two or three adjacent areas. Deep specialists remain valuable where the depth is genuinely the constraint, and less so where it is not.

What this means if you are hiring into it

Rewrite the specification from the work as it is actually done now, not from the archived requisition. The practical test is to ask the team what they spend their week on today and compare that to the document. If the job description describes production that has already been automated, it will not attract the person who can do the job that remains.

Expect the seat to be more senior and more expensive than the one it replaces, and expect fewer of them. A team that used to need four people at one level and one above them frequently now needs two and two. The total cost is not always lower, and framing the hire as a cost saving sets up a disappointment.

Interview for the failure mode rather than the tooling. A useful question in almost any function: describe a time you received work that looked correct and was not, and how you noticed. Candidates who have genuinely operated at the accountable level answer that easily and specifically. Candidates who have not tend to answer in principle.

The market this creates

The consequence for anyone hiring is a market that looks slow in aggregate and is intensely competitive at the top of it. Volume hiring in several functions is genuinely down. Competition for people who combine domain judgment with comfort operating a modern toolset is as sharp as it has been, and those people are rarely applying to anything, because they are employed, productive, and being retained deliberately.

That is the practical reason these searches so often have to be run rather than posted. The population is small, it is not in an applicant pool, and the specification is unusual enough that the people who fit it do not recognise themselves in a job advertisement.

Ready to fill a hard role?

Tell us about the search. We come back with a market read and a plan.