Fractional AI Orchestrator: The Role Your Org Chart Doesn't Have
Hosts: Trisha Merriam, Chris Carolan, Erin Wiggers
September 9, 2026
Erin Wiggers has been sitting in a role no org chart has — fractional AI orchestrator for one of her clients — and she walks the room through what the job actually turned out to be. Her answer is not a technology answer: the work is earning enough trust that a team will reach for AI at all, and step one was asking every person what they think AI is good at and what they think they themselves are good at. Trisha Merriam's check across every client portal she supports: not one has named an AI leader.
Moments from this episode
19 short cuts from this conversation.
Chris CarolanAI scales whatever it is
Chris CarolanCaring is not a business muscle
Trisha MerriamFlying blind across every portal
Erin WiggersFriction in all directions
Chris CarolanHow technical does an orchestrator need to be: and the AI committee that never changed the model
Erin WiggersHow you measure a role whose output is other people's work getting easier
Key takeaways
Trisha Merriam's own portfolio is the evidence for the whole episode: across every client portal she supports, not one has identified an AI leader — "they're all winging it, they're all flying blind."
Where a company has named someone, the job is usually governance and security. Erin Wiggers's distinction: that is AI management, not AI strategy — "literally just controls."
Nobody hired for this seat. Erin's engagement began as solutions architecture and technical HubSpot build work and became the orchestrator role through the conversations about what she had built.
Two things qualify someone to hold the seat: a deep technical understanding of how AI systems are actually built — memory, data structure, where they break — and an opinion. Having already hit the wall is what makes you the strategic play.
The stance that gets adoption is a refusal, not a sell: "I'm not trying to make AI a job on top of the job that you already have that they're not going to pay you anymore for."
The role points in every direction at once — reducing the friction of the AI decision for the team, for leadership, for clients, and in how people interact with the AI itself.
Step one of building an AI architecture is not architecture. It is establishing the company's point of view on AI, and asking each person what they think AI is good at and what they think they are good at.
Trust is the commercial constraint, not a soft one: nobody will put their name on something that feels like a robot wrote it, and in professional services that will bulldoze goodwill faster than almost anything.
When a team is not using AI, the useful question is why, not what. Not using it is a choice, and the choice is more informative than any capability survey.
Erin tracks two things: daily active users of the system, and self-reported time saved rather than machine-measured hours — because at this stage what matters is whether the work feels easier. The capability question underneath both: did you do anything this week that you otherwise would not have had time for?
A top-down process audit reads as a mandate even when it is not one. Pieces of the system remain deliberately switched off until trust catches up — and the manual step being automated may be where somebody got their best ideas.
More AI bought time to standardise, and standardising bought more time again — the compounding Erin did not plan for and now treats as the main event of the next phase.
The signal that the role is working is that it stops needing the person in it: an early adopter started running her own one-on-ones to show peers how she uses the system, and Erin found out afterwards.
Naming an AI role is not the same as filling it well. Chris Carolan's counter-example: an AI committee of a dozen people where nobody had ever tried changing the model in their Copilot conversation.
The industry keeps inventing titles for the human half of this — "forward deployed engineer," or as Erin put it, "the first interpersonal skills like engineer." None of them let you skip building a culture where people trust each other, because AI scales whatever is already there.
Keep learning
You just watched the work. Now build it.
The next step is doing it yourself — live, with people on the same path.
Get the next conversation in your inbox
One signal-dense note when a new episode lands. No noise.