Dr. Jonah Tebaa, the Lebanon-based AI strategist and author of Applied AI for Future Ready Organizations, argues that the single most consequential decision a team makes when adopting an AI system isn't which vendor to buy or which model to run. It's whether the system gets a job description.
The claim grew out of a recent engagement with a nine-person logistics operation that had already put two AI systems into production — one assigning carrier loads, one drafting responses to shipment delays — before anyone had settled who, exactly, those systems answered to. In his account of the work, Dr. Jonah Tebaa describes asking the founder a simple question: what's this AI's job title? Not a description of its function, but a title — the kind of thing that would sit on an org chart next to a name, a manager, and a set of duties someone could be held to. The founder didn't have an answer. Neither did his team.
Why "It's Just a Tool" Misses the Point
In his work, Dr. Jonah Tebaa pushes back on the instinct to treat AI systems as software rather than roles. That framing holds up for tools that produce a fixed output from a fixed input — a spreadsheet macro, a routing algorithm. It breaks down once a system is making judgment calls that used to belong, explicitly, to a person with a job description and an annual review. The moment an AI system starts deciding which carrier gets a shipment or what tone a customer-facing apology should take, he argues, it has effectively absorbed a role. Deploying it without the structure that came with that role doesn't simplify the organization — it deletes the accountability that made the work legible in the first place.
The distinction he draws is a structural one, not a trust question. It isn't about how much an organization should rely on an AI's output. It's about where the function sits, who owns what it produces, and what happens on a fixed date when someone reviews it. Get the placement right, in his view, and the trust question shrinks on its own, because there is now a specific person and a specific date attached to every decision the system makes.
The Six-Field Employee File
The framework Dr. Jonah Tebaa applied with the logistics client treats an AI system the way an HR department would treat a new hire, using six fields:
- A job title describing the function performed, not the product purchased.
- A reporting line naming the one human who owns the system's output and answers for it.
- A scope of authority defined as explicit thresholds — what the system may decide alone, and what must escalate.
- An onboarding checklist of edge cases and historical scenarios the system must clear before going live.
- A review cadence with a fixed date and a scorecard, not a vague promise to "check in."
- Retirement criteria specifying, in advance, the conditions under which the role is reassigned, retrained, or shut down.
Notably, Dr. Jonah Tebaa observes that almost none of this information was new to the founder — he already held most of these judgments informally. The value of the exercise, in his telling, wasn't discovery. It was making the founder's existing judgment explicit enough to be enforced, questioned, and reviewed by someone other than the founder himself.
What Changed Once the Roles Were Written Down
According to Dr. Jonah Tebaa, the clearest evidence of the framework's effect showed up in how the operations team talked about the systems' mistakes. Before the exercise, a bad outcome was described vaguely — "the AI did something weird" — a sentence with no subject and no owner. Afterward, the same event became "the Carrier Assignment Analyst flagged three shipments outside its threshold," a sentence with a named role, a scope, and an implicit next step.
He also points to a second, less obvious shift: escalation habits changed from group speculation to a direct route to the named manager who had agreed, in writing, to own decisions in that band. And a third shift, which he argues founders underestimate — a good week from the AI system became attributable to somebody, visible on a scheduled review rather than absorbed as ambient good luck. In Dr. Jonah Tebaa's framing, structure isn't only a tool for catching failure faster. It's what makes good performance legible enough to be recognized at all.
The broader argument he's making to business leaders is that an AI system without a job title effectively has no manager — and a system with no manager becomes everyone's problem the moment it fails, and no one's credit the moment it works. For teams running AI in production without having answered that question, Dr. Jonah Tebaa's suggestion is straightforward: write the job description before debating how much to trust the system. The trust question tends to resolve itself once the role is on paper.