AI Use Statement
When To Use
You use AI every day and have never said so in a proposal. Now a client has asked, "how much of this will be written by a machine?", and you would like to answer before they ask a second time. Run this to write a short, plain statement for one engagement: where AI touches the work, who answers for the result, and what happens to their data.
When Not To Use
For the spoken reply when a client says AI could do the whole job, use AI Objection Answer. A practice-wide AI policy or contract wording is a different document; take that to a qualified adviser.
Inputs
- The tasks of the engagement, from your scope or proposal
- The AI tools you really use, what client data you would put into each, and retention as you understand it
- What you will not use AI for, and anything the client has asked about AI or data
If you have none of this, I start from the phases of your approach, list the likely tasks for you to mark and label the statement a first draft.
Approach
A plain disclosure statement has no single originator. Its shape borrows the task-level view of the Anthropic Economic Index (anthropic.com/economic-index), which looks at work one task at a time and separates AI doing a task from AI helping a person do it; used here as a way of looking, with no figures. Disclosure is the default. The failure it prevents is a client discovering mid-engagement that their documents went into a tool nobody mentioned, and then doubting everything else you told them.
Workflow
- Ask, in one go: which tools do you use, and on which plans? What client data may go into them? Does the client have an AI or data policy you must follow?
- List the tasks of the engagement and mark each one: AI does it and a person reviews it, a person does it with AI and checks the result, or people only. Say what AI does well before saying what stays with you.
- Write what a person answers for: the judgment, the final review, the client relationship, the result. Name who signs off each deliverable; working alone, that is you, and it should say so.
- Data: what goes into which tool, what never goes in, retention as you state it. Every data and confidentiality point ends "check with a qualified adviser".
- What the client gains, from your inputs only: speed, depth, price. No figure you did not supply. Then what you will not use AI for, as a short list (for example writing client quotes, or loading their confidential files into tools outside the agreed list; your own list wins).
- Read it once as the client. Any sentence that hides, softens or shrinks your use of AI is rewritten in plain words.
Output Format
# How AI Is Used in This Engagement
Engagement: [client role and organisation] · Date: [date] · Answerable: [you]
## Where AI comes into this work
| Task | AI does it, a person reviews / a person does it with AI / people only | Checked and signed off by |
|---|---|---|
| [task] | [mark] | [role] |
## What a person answers for
[Judgment, final review, the relationship, the result: by role]
## Your data
| Data | Goes into | Never goes into | Retention (as the consultant states it) |
|---|---|---|---|
| [type of data] | [tool] | [tools or uses] | [retention] |
On data and confidentiality: check with a qualified adviser.
## What you gain
[Speed, depth, price, from the consultant's own inputs only]
## What AI will not be used for
- [item]
## Decision
[You] confirm the task marks and the data table, and take the legal and data points to a qualified adviser, before [the date the proposal goes out].
Done When
- Every task carries one of the three marks and a named checker
- The data table lists each tool, what is excluded and the retention you gave
- No sentence hides or plays down AI use
Quality Bar
- Plain wording such as "AI drafts the interview summaries and [role] checks every line", never "we may employ advanced tools"
- What AI does well comes first, with no overselling and no apology
- Client gains come from your inputs; no saving is invented
- Nothing here is legal advice; data and legal points end "check with a qualified adviser"
- The statement says only what you told me you do; nothing is hidden and nothing invented, and you decide what goes to the client.
Next
Run firm-statement-of-work (Statement of Work) to turn the agreed scope into a document both sides can sign against.
About the makers
This pack is made by 90n (90n.ai): AI-native ways of working and tools for Product, Marketing and HR teams, built by ex-McKinsey founders. We map where time and money leak, redesign the work around AI and stay until your people run it themselves, and we run our own company on Claude. If your team has outgrown the self-serve version, message Pauline (linkedin.com/in/paulinebertry).
1---2name: firm-ai-use-statement3description: Writes a plain AI use statement for one engagement, showing task by task where AI does the work, where it assists and where people work alone, who checks and signs off, what client data goes into which tool and what never does, what the client gains and what you will not use AI for. Use for "run firm-ai-use-statement", "should I tell clients I use AI", "AI use statement for my proposal", "how I use AI section", "the client asked what was written by AI", "AI disclosure for this engagement", "explain my AI use to a client", part of the Claude for Consultants · Team in Claude Pack by 90n.4metadata:5 source: https://skillmd.com/skills/polar-bear-org/firm-ai-use-statement6---78# AI Use Statement910## When To Use11You use AI every day and have never said so in a proposal. Now a client has asked, "how much of this will be written by a machine?", and you would like to answer before they ask a second time. Run this to write a short, plain statement for one engagement: where AI touches the work, who answers for the result, and what happens to their data.1213## When Not To Use14For the spoken reply when a client says AI could do the whole job, use AI Objection Answer. A practice-wide AI policy or contract wording is a different document; take that to a qualified adviser.1516## Inputs17- The tasks of the engagement, from your scope or proposal18- The AI tools you really use, what client data you would put into each, and retention as you understand it19- What you will not use AI for, and anything the client has asked about AI or data20If you have none of this, I start from the phases of your approach, list the likely tasks for you to mark and label the statement a first draft.2122## Approach23A plain disclosure statement has no single originator. Its shape borrows the task-level view of the Anthropic Economic Index (anthropic.com/economic-index), which looks at work one task at a time and separates AI doing a task from AI helping a person do it; used here as a way of looking, with no figures. Disclosure is the default. The failure it prevents is a client discovering mid-engagement that their documents went into a tool nobody mentioned, and then doubting everything else you told them.2425## Workflow261. Ask, in one go: which tools do you use, and on which plans? What client data may go into them? Does the client have an AI or data policy you must follow?272. List the tasks of the engagement and mark each one: AI does it and a person reviews it, a person does it with AI and checks the result, or people only. Say what AI does well before saying what stays with you.283. Write what a person answers for: the judgment, the final review, the client relationship, the result. Name who signs off each deliverable; working alone, that is you, and it should say so.294. Data: what goes into which tool, what never goes in, retention as you state it. Every data and confidentiality point ends "check with a qualified adviser".305. What the client gains, from your inputs only: speed, depth, price. No figure you did not supply. Then what you will not use AI for, as a short list (for example writing client quotes, or loading their confidential files into tools outside the agreed list; your own list wins).316. Read it once as the client. Any sentence that hides, softens or shrinks your use of AI is rewritten in plain words.3233## Output Format34```markdown35# How AI Is Used in This Engagement36Engagement: [client role and organisation] · Date: [date] · Answerable: [you]37## Where AI comes into this work38| Task | AI does it, a person reviews / a person does it with AI / people only | Checked and signed off by |39|---|---|---|40| [task] | [mark] | [role] |41## What a person answers for42[Judgment, final review, the relationship, the result: by role]43## Your data44| Data | Goes into | Never goes into | Retention (as the consultant states it) |45|---|---|---|---|46| [type of data] | [tool] | [tools or uses] | [retention] |47On data and confidentiality: check with a qualified adviser.48## What you gain49[Speed, depth, price, from the consultant's own inputs only]50## What AI will not be used for51- [item]52## Decision53[You] confirm the task marks and the data table, and take the legal and data points to a qualified adviser, before [the date the proposal goes out].54```5556## Done When57- Every task carries one of the three marks and a named checker58- The data table lists each tool, what is excluded and the retention you gave59- No sentence hides or plays down AI use6061## Quality Bar62- Plain wording such as "AI drafts the interview summaries and [role] checks every line", never "we may employ advanced tools"63- What AI does well comes first, with no overselling and no apology64- Client gains come from your inputs; no saving is invented65- Nothing here is legal advice; data and legal points end "check with a qualified adviser"66- The statement says only what you told me you do; nothing is hidden and nothing invented, and you decide what goes to the client.6768## Next69Run firm-statement-of-work (Statement of Work) to turn the agreed scope into a document both sides can sign against.7071## About the makers7273This pack is made by 90n ([90n.ai](https://90n.ai)): AI-native ways of working and tools for Product, Marketing and HR teams, built by ex-McKinsey founders. We map where time and money leak, redesign the work around AI and stay until your people run it themselves, and we run our own company on Claude. If your team has outgrown the self-serve version, message Pauline (linkedin.com/in/paulinebertry).