ChamberOne

When clients ask about your AI use and your fees

By the ChamberOne team 4 minute read

Corporate clients have started to put questions about artificial intelligence into their outside counsel guidelines and panel questionnaires. Some ask whether you use it. Some ask you not to. Some ask whether your bills will fall. A vague reply written in a hurry creates trouble later, because it becomes a promise. This article helps partners decide a position in advance and write answers that are honest, specific and consistent across clients.

Key points

  • Clients usually worry about confidentiality, quality, cost and control; answer those worries, not just the wording.
  • Agree a one-page internal position on AI before the questionnaires arrive.
  • Be specific about data handling, cite vendor statements as theirs, and record any client prohibition on the matter file.
  • On fees, avoid promising discounts you have not modelled and avoid billing as if the work was done by hand.

What clients are really asking

Behind the wording of a questionnaire there are usually four worries. First, confidentiality: will our information be fed into a public tool? Second, quality: will an unchecked machine produce the advice? Third, cost: if work is quicker, why is the bill not smaller? Fourth, control: will we know when AI was used on our matter?

Answering the worry, not just the question, produces a better reply. A bare 'yes, we use AI' invites alarm. 'Yes, in these ways, with these controls, and a lawyer reviews every output' answers the worry underneath.

Decide your position before the questionnaire arrives

Firms that answer these questions well have usually agreed an internal position first. It does not need to be long. A single page covering the points below is enough, approved by the partners and reviewed when your tools change.

  • Which tools the firm uses, which it does not, and who approves new ones.
  • What kinds of client data may go into which tool, and what may never be entered.
  • Where AI is used on client work and where it is not.
  • How outputs are reviewed and who is responsible for them.
  • Whether and how clients are told when AI is used on their matter.
  • How the firm treats AI-related cost in its fees.

Answering the data and confidentiality questions

This is the section clients read most closely. Be specific about what you know and honest about what you do not. Describe the controls you operate: which providers receive text, whether it is kept or used for training under your agreement, where it is processed, and who within the firm can use the tool.

Do not copy a vendor's marketing line into your answer. If you are asked a detailed question about the vendor, get the answer from the vendor in writing and cite it as theirs. Overstating what a supplier does is one of the easiest ways to make a promise you cannot keep.

If a client prohibits AI on its matters, record that instruction on the matter file and make sure the team can see it. A prohibition that sits in an email in one partner's inbox will be broken by someone who never saw it.

Answering the fee questions

A client who asks whether AI will lower costs is entitled to an honest conversation. Faster drafting or review may reduce hours on some tasks. Other tasks do not shrink at all, and checking AI output takes time that did not exist before. The truthful answer is usually that it depends on the task, and that you will discuss it matter by matter.

Be careful about two traps. One is promising a discount you have not modelled. The other is billing the client as though the work took the time it would have taken by hand. Think about how your fee arrangements treat efficiency: time-based billing rewards slowness and fixed fees reward speed. Many firms find that a conversation about scope and fee structure at the start of a matter serves both sides better than an argument over an invoice at the end.

Also decide whether the cost of an AI tool is overhead or a disbursement. Check your engagement terms and the guidance from your regulator or bar before charging a client separately for software.

Quality, review and accountability

Clients want to hear that a named lawyer stands behind every piece of advice. Say so plainly, and say what review means in your firm: who reads the output, what they check, and how citations and figures are verified. Avoid language that suggests the machine is the author.

If a client asks for an audit right or a record of AI use on their matter, find out what your tools can actually produce before you agree. A record of which model was used and when is something some systems provide and others do not.

Drafting the answers

A few habits keep replies safe and useful.

  • Write in plain language and avoid adjectives that cannot be tested, such as secure or industry-leading.
  • State what you do, then what you do not, then what you will check with the client first.
  • Keep a library of approved answers and update it, so two partners do not give two different replies to the same client.
  • Date each answer. A statement about your tools is true on a particular day.
  • Route unusual questions to one person, so contradictions are caught.
Do
  • Write in plain language
  • State what you do, then what you do not, then what you will check
  • Keep a library of approved answers
  • Date each answer
  • Route unusual questions to one person
Don't
  • Use adjectives that cannot be tested, such as secure or industry-leading
  • Copy a vendor's marketing line into your answer
Writing safe replies to client questions

What to do when you have no good answer yet

Sometimes the honest reply is that the firm has not settled its position. Say so, and say when you will. A client will usually prefer a short, candid answer with a date attached to a confident paragraph that later proves wrong. If a question is about a tool you are only trialling, describe the trial as a trial and say what data is excluded from it. Then follow through, and update the client when the position changes.

Treating the questionnaire as a conversation

The best outcome of a procurement exercise is not a ticked box but a shared understanding. Offer a short call with the client's legal operations or procurement team to walk through your position. Ask what outcomes they care about. Some will want faster turnaround, some will want predictability, some will want reassurance on data. Your AI position is easier to defend when it is tied to something the client values.

This article is general information about the law, legal practice and technology, current when it was published. It is not legal advice. Laws and guidance change: check the sources linked above and take advice on your own circumstances from a lawyer qualified where you are.

Read next

Outside counsel guidelines and AI: answering clients