Corporate clients are no longer waiting for law firms to decide how AI should affect legal bills.
They are writing the rules themselves.
Outside counsel guidelines have always covered rates, staffing, expenses, time-entry detail, and invoice review. AI adds a new set of questions: Was an AI tool used? Was client data entered into it? Who reviewed the output? Did the firm charge for the tool, the lawyer’s time, or both? Did the efficiency show up in the invoice?
For firms, this is not only a billing issue. It is a client governance issue that touches security, supervision, workflow design, and commercial strategy at the same time.
The signal from clients is getting clearer
In May 2026, Meta’s global head of legal operations, Mike Haven, argued publicly that hourly billing would become the exception within five years. His point was not that firms should avoid AI. It was that a pricing model built on elapsed time becomes harder to defend as machines compress more of the work.
Zscaler’s published Outside Counsel Billing Guidelines are more concrete. The company encourages appropriate AI use, requires safeguards for confidential information, prohibits using Zscaler data to train AI tools, requires human attorney review, and says time and cost associated with AI-generated work product may not be passed on to the company. A discrete use of generative AI must also be noted in the time entry.
One public document does not establish a universal rule. It does show what a sophisticated client can ask a firm to prove.
The Association of Corporate Counsel now gives in-house teams a set of questions for evaluating outside counsel’s AI readiness. The themes are consistent: policy, accountability, client choice, data exclusion from training, quality controls, and billing transparency.
If a client has to ask how your firm uses AI, the answer should already exist in writing.
The disclosure trap under hourly billing
Hourly firms can end up in an awkward position.
If a lawyer discloses that AI reduced a task from six hours to two, the client may reasonably expect a smaller invoice. If the firm does not disclose the use of AI and the client later discovers it, the firm may face a billing dispute, a trust problem, or both. If the firm bills the historical time instead of the actual time, it also creates an ethics problem.
The issue is not solved by adding a vague “technology fee.” A client needs to understand the basis of any charge, and a firm needs to show that the fee is permitted by the engagement and the client’s guidelines.
Value-based, fixed, capped, or phased fees can reduce this conflict because the parties agree on the service and price before delivery. Efficiency can improve the firm’s margin without creating a fictional time entry. The structure still needs reasonable fees, clear scope, and transparent client consent.
Audit the work before the client does
Start with the firm’s ten largest corporate clients. For each one, gather the current engagement letter, outside counsel guidelines, rate agreement, billing instructions, information-security addendum, and any AI-specific correspondence.
Then map the services delivered to that client against work that is already being accelerated by AI.
Common areas include:
- first-pass legal and factual research
- document and testimony summaries
- contract clause extraction
- chronology and timeline construction
- routine correspondence
- first drafts built from approved precedent
- invoice review and narrative cleanup
For each task, answer five questions:
- Does the client permit AI use for this work?
- Is disclosure or advance approval required?
- Can client data enter the selected tool under the contract?
- What human review is required and who performs it?
- How must the time and technology cost appear on the invoice?
This turns a general AI policy into a matter-level operating rule.
Build an OCG control layer
Most firms store outside counsel guidelines as PDFs in a document system or billing folder. That is necessary, but it is not enough.
The people doing the work need the relevant rules at the point where they act. The billing team needs them when it reviews time. The AI workflow needs them before data is sent to a model. The responsible lawyer needs them before work product reaches the client.
A useful control layer translates the client’s document into structured rules:
- approved and prohibited tools
- data classification and residency restrictions
- disclosure and consent requirements
- required human review
- prohibited charges
- time-entry wording
- staffing limits
- exception and approval contacts
Those rules can then be attached to the matter record and checked at three points.
Before the work
Confirm that the tool and data are permitted. If approval is required, capture it before the prompt is sent.
Before the output leaves the firm
Confirm that a qualified lawyer reviewed the substance, citations, and client-specific requirements. Save the review record with the matter.
Before the invoice leaves the firm
Check time narratives, AI disclosures, prohibited costs, and fee structure against the client’s current rules.
This is where AI can become a control rather than only a risk. A matter-aware review system can flag invoice entries that conflict with the client’s guidelines before the client finds them.
Give clients a coherent answer
Firms often answer AI diligence questions one department at a time. IT describes security. Innovation describes tools. Risk describes policy. Billing describes time entries. The client receives four partial answers and has to decide whether they fit together.
A stronger response is one operating story:
- Purpose: the firm uses AI for defined tasks where it improves speed or consistency.
- Tooling: only approved tools and accounts may process client work.
- Data: client information is handled according to the engagement, vendor contract, and matter restrictions.
- Review: a responsible lawyer verifies the output and remains accountable for the work.
- Billing: time and costs are recorded according to the engagement and the client’s guidelines.
- Evidence: approvals, prompts where appropriate, outputs, review steps, and invoice checks are logged.
That story is more credible when it is backed by system behavior. A policy that says lawyers must use approved tools is weak if anyone can paste client data into an unapproved account. A disclosure rule is weak if the billing system has no way to record it.
Turn client pressure into a sales advantage
AI billing guidelines do not have to be treated as defensive paperwork.
A firm that arrives at a panel review with a clear AI and pricing model can make the client’s job easier. It can explain which tasks are accelerated, where human judgment remains essential, how data is protected, and how efficiency affects the fee. That is a stronger proposition than silence or a generic promise to “use AI responsibly.”
The client is asking for predictability and control. A firm that can demonstrate both has something concrete to sell.
What to do this month
- Collect the current billing and AI instructions for the firm’s largest clients.
- Identify the five services most exposed to AI-driven time compression.
- Map each service to disclosure, data, review, and billing requirements.
- Add those rules to matter intake, work review, and pre-bill review.
- Prepare a client-facing explanation of the firm’s AI operating model.
- Test one value-based or fixed-fee proposal where scope is stable.
Waiting for a rejected invoice is the expensive way to learn what a client expects.
Need to make client guidelines operational?
Jinka helps firms turn billing, AI, and security requirements into matter-aware workflows that people can actually follow.
Sources
- Zscaler: Outside Counsel Billing Guidelines
- Association of Corporate Counsel: GenAI transparency and readiness questions for outside counsel
- Association of Corporate Counsel: Generative AI’s growing strategic value for corporate law departments
- Law.com: Meta legal ops chief on the billable hour
- ABA Formal Opinion 512