Operator Notes · The AI, in production

Every Role Moves Up an Altitude.

Andrew Hahm · June 2026 · 3 minute read
WHERE THE WORK WAS WHERE IT GOES Elevate, don't automate. ah.

The fastest way to kill an AI rollout on a briefing team is to let it sound like an automation play. The failure is predictable. The team hears "agents," translates it to "headcount," and quietly stops feeding the system the corrections that would have made it good.

So I start from a different claim, and I mean it literally. Every role on the team moves up an altitude. None disappears. Each one sheds its lowest-value work.

RoleShedsPicks up
Specialist or coordinatorData entry, form chasing, assembling recaps by hand.Running the intake, recap and survey workflows, and owning the exceptions.
Engagement managerHours of research and first drafts.Certifying the research, shaping the room, time with the customer.
ManagerStatus spreadsheets and manual capacity tracking.Managing the system: what is blocked, what is drifting, where to coach.
DirectorReporting from anecdotes.Strategy from a real record of what customers said.
The engagement manager is where AI matters most. Every briefing is bespoke, and the hours are the longest.

From producing to certifying

The operators who do well in this model stop being producers and become editors and certifiers. The scarce skills are judgment (is this the right frame for this room?), taste (does this clear the bar?), and fact-certification (is this true?).

That is a more senior version of the job. It is also the truthful answer to the question people are afraid to ask out loud. This takes the parts of the job nobody went into the field to do.

Adoption is structural, not motivational

A town hall will not fix the fear. Three design choices will.

Lead with the guardrail. The first thing the team hears is that the agent is never the last step, and that they certify everything that reaches an executive.

Pick the right first workflow. Choose the one that eats the most tedious manual time, so the team's first experience of AI is relief.

Make people the visible owners of quality. Their names are on the certifications. Their corrections improve the system. Their judgment is the standard everything else is held to.

What stays human

What AI can't do is the part customers remember. The warmth in the room. The read on an executive who has gone quiet. The decision to drop the agenda and follow the conversation. AI prepares the experience. People deliver it, and the point of the system is to buy back the hours to deliver more of it.

The team that adopts fastest is the one that experiences the system as getting their afternoons back.

Always forward

If you run customer programs or executive engagement and you are working through this inside your own team, I'd like to compare notes.