What is AI-Assisted Billing?

Why It Matters

The billable hour converts time into revenue, and a tool that compresses time compresses revenue unless the firm changes what it charges for. That is the economics of the engagement rather than the tooling, and clients have noticed: a large share now expect a lower fee when AI was involved, and some corporate legal departments have written AI billing restrictions directly into their outside counsel guidelines.

For a firm, the exposure runs in two directions. Billing the eliminated hours is a fee problem under the professional rules. Failing to capture the efficiency at all leaves the saving with the client by default, which is a margin problem with no rule against it and no upside either.

What the Rules Allow

The ABA’s guidance is specific on two points. A firm may not bill for hours a tool eliminated, and may not charge for the time spent learning a tool unless the client specifically requested that tool for that matter. Where the firm uses a costly proprietary platform, the cost can be treated as office overhead or passed through in some proportion, provided the client is told in advance and consents.

Nothing in that prevents a firm from earning more on the same engagement. It prevents the firm from earning it through a time entry that no longer describes the work.

Where It Breaks

The compliance artifact is the time record, and most firms cannot produce one that separates AI-assisted work from professional work. Without that separation, a firm cannot show it complied with the fee rule, cannot answer a client asking where the saving went, and cannot price the next engagement from real effort data.

The second break is realization. Discounting the AI-assisted portion without renegotiating the fee structure erodes margin quietly, engagement by engagement, which is the pattern behind the reported gap between firms that have moved to value-based pricing and firms that have not.

The third is the trap in a hybrid engagement: if the AI-assisted bucket is priced at a flat rate and the professional bucket by the hour, the incentive to move work between buckets is real and needs an owner.

How Flytebit Handles It

The record separates AI-assisted time from professional time per deliverable, so the fee basis is provable rather than reconstructed. The same decision record that carries the sources and the signer carries the effort data, which means one artifact answers the ethics question and the pricing question together. Cost attribution runs underneath it for the tool spend itself. The industry application is on our Professional Services page, and the operating model is our LLMOps work.