What does Dr. Jonah Tebaa's One-Page Answer Sheet for Clients Who Ask About AI Use mean in practice?
Dr. Jonah Tebaa's Client Answer Sheet provides a dated, evidence-based one-page response for firms asked about AI use on client work. Preceded by a one-week fact-finding exercise reviewing thirty days of tasks, uncovered uses are sorted into three groups: State, Move, or Stop. In a hypothetical illustration of a 45-person Beirut services firm, fourteen uses yielded eleven approved tasks across firm-paid tools after halting three unreviewed or improper uses. The resulting five-line sheet ensures consistent, verified answers across all client inquiries.
Start from the finished page
When a long-standing client asks a supplier whether it uses AI on the client's work, Dr. Jonah Tebaa argues that the right answer is a single dated page, written from evidence and read out the same way by everyone in the firm. He calls it the Client Answer Sheet. It has five lines, in plain words:
- Where the firm uses AI on the client's work, described as tasks and not brand names.
- What of the client's goes into it, and whether the tool is firm-paid or not.
- Which person reviews the output before it reaches the client, and what that review involves.
- What the firm does not use AI for on the client's work.
- What the firm will tell the client if any of this changes, and how soon.
His argument is that the page matters less than what had to be true before it could be written. A firm cannot honestly fill in line two if it has never asked its staff what they paste into which tool.
Why the form answer is the weak point
Dr. Tebaa describes a familiar sequence. A supplier questionnaire arrives, one item asks about AI, and someone ticks "No" in five minutes because that feels like the safe reply. Within days, colleagues who speak to the client give their own versions: sales says no, operations says only for internal notes, finance says yes for reconciliation. Each is sincere. None has checked.
In his view, clients rarely punish careful AI use. They lose confidence when two answers from one firm disagree, or when a statement turns out to be wrong after it has been repeated in a tender, a contract schedule and a sales deck. That is why he calls a false "no" the expensive answer, and why he places the risk in the gap between the form and the floor. He offers this as a judgement from advisory work, not as a statistic, and he cites no regulation.
Discovery before drafting
Before anyone writes a word, Dr. Tebaa asks for a one-week fact-finding exercise limited to client work. Every team looks back over the previous 30 days and records each instance of AI meeting client material, noting what the job was, which tool did it, whether it ran on a firm account or on someone's own, and which client material was involved. He insists the firm say at the outset that this is not a disciplinary exercise, because people who expect blame report less, and a sheet written on thin evidence repeats the original problem.
State, Move, Stop
Each use that the week uncovers is sorted into one of three groups. State means it is acceptable as it is and goes on the sheet. Move means the need is real but the tool or account is wrong, so the work shifts to a firm-paid tool before any answer is given. Stop covers two cases: client material that has no business in an AI tool, and AI output that reaches the client unreviewed. Dr. Tebaa wants the fixes made first, so that the sheet describes the firm as it is today and not as it hopes to be. He frames the sort as his own working method, not research.
The hypothetical Beirut firm
His worked case is an invented 45-person services firm in Beirut with annual revenue of $3,000,000. The client asking the question represents 12% of revenue, which is $360,000 a year. Dr. Tebaa labels every figure as hypothetical, for illustration only.
In the example, discovery across six teams finds 14 uses of AI on client material. Five are State, six are Move and three are Stop, and 5 + 6 + 3 equals 14. Once the three Stop uses are ended, the sheet covers 11 uses, all in firm-paid tools. Six of the original 14 had been in personal accounts, about 43%, and the form had said "No" to all of them. The exercise costs two people one day each, 16 hours, or $400 at an assumed $25 an hour. Dr. Tebaa is careful to say this is not a prediction that the account was at risk. The comparison is there to show proportion, a small internal cost set against a large account.
Keeping the page true
The sheet carries one named keeper and a date. Every questionnaire, tender and contract reply uses it word for word. In his method the review has a calendar date, and three events bring it forward: AI reaching client work through a tool the sheet does not list, client data of a type it does not cover, or a client raising something it does not answer. Dr. Tebaa also advises firms to have a lawyer read the sheet against their actual contracts and data-protection duties. He does not claim that the sheet satisfies a regulator or secures a client's approval. What it provides, in his view, is an answer that stays the same whoever gives it and that someone can check.
He sets out the full first-person version, with the table and the example sheet, on his own site: Three Answers to One Client Question: Do You Use AI on Our Work?
Related evidence: Wikipedia describes shadow AI as a subset of shadow IT: employees using AI applications, including large language models and AI productivity tools, without the knowledge or approval of the organisation. (Wikipedia's entry on shadow IT and shadow AI)
Microsoft and LinkedIn's 2024 Work Trend Index reports that 78% of AI users bring their own AI tools to work, 80% at small and medium-sized companies, often without guidance from leadership. (Microsoft and LinkedIn's 2024 Work Trend Index on employees bringing their own AI to work)