AI for Lawyer Billing: Draft Narratives From Real Work

Use AI for lawyer billing to draft descriptions from verified work records, then have the responsible lawyer check each entry before it reaches an invoice. Keep actual time in the firm’s approved timekeeping system; an AI draft must not invent work, reconstruct unsupported hours, or decide what the client owes.

A record such as fenbs can preserve internal task history and approved review rules, without becoming your billing ledger. Start with the evidence for the work, use an approved tool, and keep the final billing decision with a person.

How do you draft a billing narrative from verified work?

1. Collect the approved time entry, date, matter identifier, work description, and supporting record. Resolve missing or conflicting information before generating text.
2. Check whether the chosen AI tool is approved for the information involved. Remove unnecessary details and follow the firm’s confidentiality and client instructions.
3. Ask for a concise description of the work actually recorded. Require the model to preserve the supplied facts and flag gaps rather than fill them.
4. Compare the draft with the original record and the client’s billing guidelines. Check the action, purpose, date, matter, and any task code.
5. Have the responsible lawyer approve the narrative and any billing treatment. Transfer only the reviewed entry into the billing workflow and retain the supporting record.

Start with a small set of completed entries. Review them individually before allowing the same prompt to process a longer set. A consistent writing style helps only after you have established that the output preserves the facts.

What do US ethics rules mean for AI-assisted billing?

The ABA’s Formal Opinion 512↗ (opens in a new tab) explains that hourly billing must reflect the lawyer’s actual time, even when AI makes the work faster. It also addresses review time, charges for AI tools, and the limits on charging clients for learning routinely used technology.

The opinion is guidance based on the ABA Model Rules. Check your jurisdiction’s adopted rules, applicable ethics opinions, the engagement agreement, and the client’s billing requirements before setting a policy. A model’s confident answer about what is billable cannot establish those obligations for your matter.

Under ABA Model Rule 1.5↗ (opens in a new tab), fees and expenses must be reasonable, and the basis or rate must be communicated as the rule provides. A polished narrative does not cure an unsupported charge. Flat or contingent arrangements also require their own review; do not automatically apply hourly assumptions to them.

For this workflow, ask AI to describe evidence rather than interpret a fee agreement. Put an unfamiliar expense, disputed time entry, or requested exception into a review queue. Decide it before the invoice is approved. Preserve the firm’s approved review instructions in fenbs Decisions and rules so connected assistants read them before helping.

What should a good narrative include?

Describe a concrete action and its purpose at the level the client’s guidelines require. A vague phrase such as worked on matter gives little basis for review. An inflated phrase that adds analysis, strategy, or finalization to a simple document check creates a different problem: it changes what happened.

These examples do not establish billability. They illustrate how to preserve the meaning of an existing record. Do not add a client name, legal issue, outcome, or duration merely because it would make an entry more informative.

Draft one billing narrative from the verified record below. Describe only the action and purpose stated in that record. Do not add work, time, people, legal conclusions, or outcomes. Do not infer duration from timestamps. Preserve supplied matter and task codes exactly. If required information is missing, return NEEDS REVIEW and name the gap. Follow the approved narrative rules supplied with the record.

How do you protect client information?

A billing entry can reveal information about a representation even when the underlying document is absent. ABA Model Rule 1.6↗ (opens in a new tab) addresses confidentiality and reasonable efforts against unauthorized disclosure or access. Treat the narrative and the material used to draft it as part of that review.

Before using a tool, understand who can access inputs, how they are retained, and whether your intended use fits the firm’s obligations. A matter nickname does not necessarily remove identifying details. If the firm has not approved the use, draft the entry within the existing approved system.

Provide the minimum description needed for the task. Do not upload a matter file when a permitted work description would suffice. Review the output for unnecessary strategy, sensitive allegations, or details unsuitable for the invoice recipient, including outside billing reviewers.

General preparation belongs in lawyer AI training; broader delegation belongs in AI agents for lawyers. This process is limited to preparing and checking billing narratives from records the firm has already verified.

Which errors should a reviewer look for?

Check substance before grammar. Models can merge neighboring entries, move work to the wrong matter, or make a preliminary review sound complete. A fluent sentence is easy to approve quickly, which makes comparison with the original entry essential.

• Invented work: Does every action in the narrative appear in the verified record?
• Unsupported time: Has the draft inferred hours from task activity, document history, or an estimate?
• Duplication: Does the same work appear twice after entries were combined or rewritten?
• Wrong scope: Does the description exaggerate completion, responsibility, or legal analysis?
• Client requirements: Do wording, grouping, and task codes follow the approved guidelines?

A task being marked complete establishes a workflow state, not the duration of legal work. A document timestamp can help locate evidence, but it does not show uninterrupted work. Where records are incomplete, ask the person who performed the work and follow the firm’s correction process.

Save recurring drafting errors as internal lessons without client details. An AI brain is useful when the next assistant needs the firm’s approved writing rules and known pitfalls. It does not make historical activity into a billable time entry. Use fenbs History to check who changed an internal task while the billing system remains the source for actual time.

How to do this with fenbs

1. Create a free account, then add a project for improving the firm’s internal narrative review process.
2. Record a person’s approved rules in Decisions and rules: draft from verified records, never infer duration, and require lawyer review before use.
3. Add sanitized tasks for process improvements and lessons from recurring errors. Keep client information and time entries in the approved matter and billing systems.
4. Connect a supported assistant through Integrations. Ask it to read the rules before helping with permitted process work and save progress before stopping.

fenbs does not track billable time, create invoices, or connect directly to a legal billing system. If you need those functions, use your firm’s approved billing software. fenbs fits the shared internal rules and follow-up work around the process.

The AI at work library and guide to using AI at work offer other reviewed workflows. For billing, begin with one verified record and one human-approved narrative before expanding the process.