Broadly, I agree with this, but the role is probably bigger than just “agent manager”.
What’s really being described here is someone deeply embedded in the work itself, reshaping it so agents can genuinely be useful.
They discover where the real work happens and rethink those workflows around what agents can actually do well, giving them the right context and tools.
Then they decide where humans still need to step in and stay accountable for whether it all works once it’s live.
The part I find most interesting is the centralisation question.
I do not think this stays as one tidy role sitting inside IT or a central AI team for very long.
It will split by department:
- Sales will need its own version
- Legal
- Finance
- Ops and support too
Each area has its own mess, its own judgement calls, and its own way of working.
A generalist will only get you so far.
At some point, you need people who understand both the domain and the wiring underneath it.
Also worth saying: the hard bit is not plugging APIs together.
The hard bit is understanding how information actually moves through a business, where an agent can be trusted, where a human needs to step in, and how to keep the whole thing steady when the models change, the data shifts, or reality gets messy.
That takes more than technical skill. It takes judgement and, most of all, taste.
The shape of the role is real.
Most companies just haven’t found the words for it yet.