Every article in this series has grown from a real conversation. This one grew from several of them, all variations on the same exchange. A client’s general counsel mentions, almost in passing, that her outside law firm has told her it uses AI. She seems reassured by that. Then I ask the follow-up question that no one had thought to ask: does your firm’s AI use translate into lower fees for you?
A pause. Then: “I honestly don’t know.”
That gap — between knowing your lawyers use AI and knowing whether you benefit from it — is where a significant amount of client money is quietly disappearing. This post is about closing that gap.
The Efficiency Problem Nobody Talks About
When law firms describe their AI capabilities, they tend to emphasize two things: speed and accuracy. AI helps us research faster. AI helps us review documents more thoroughly. AI helps us prepare more completely. All of that may be true. What goes unmentioned is the third variable: who keeps the savings.
Here is the problem in concrete terms. A task that once took fifteen lawyer hours now takes three. A lawyer who bills honestly bills three. But in many firms — and this is the part clients are not told — the lawyer still bills fifteen.
This happens for reasons that are structural, not always intentional. Many firms use what they call “value billing” — the practice of billing for the perceived value of the work rather than the time actually spent. Under that model, if legal research has historically cost fifteen hours of attorney time, it may continue to be priced at fifteen hours even after AI has compressed the actual time to three. The client sees the same line item. The hours have not changed. Only the work has.
There are other pressures that produce the same result. Associates at large firms typically operate under minimum billable hour requirements — often two thousand hours a year or more. Bonuses are tied to hours billed, not hours worked. In that environment, an associate whose AI tools have cut her research time in half faces an institutional disincentive to reflect that efficiency honestly on the invoice. The firm’s revenue model, the associate’s compensation structure, and the client’s interest in paying only for time actually spent are pointed in three different directions. The client’s interest is the one that loses.
This is the efficiency problem nobody talks about — not because it is new, but because the arrival of AI has made it dramatically more significant. The gap between time spent and time billed has always existed at the margins. AI has made that gap structural, and large, and growing.
What the Levels Actually Mean for Your Invoice
In an earlier post in this series, we described four levels of AI sophistication in legal practice: Foundation, Intermediate, Advanced, and Expert. Understanding those levels matters not just for evaluating your lawyers’ capabilities, but for understanding what you should expect on your invoice.
At the Foundation and Intermediate levels — where most lawyers who use AI at all are operating — AI functions as a fast and capable assistant. It performs initial document review, drafts motions and correspondence, and handles research that would otherwise fall to associates. Those tasks used to generate substantial associate hours on client invoices. They no longer need to, and the client’s bill should reflect that.
At the Advanced and Expert levels, AI is doing work that previously required significant senior attorney time, or that did not exist at any price: comprehensive judicial research, expert witness cross-preparation built from complete prior testimony, AI-driven jury profiling, settlement valuation modeling. At these levels, the question is not just whether you are being billed less — it is whether the capabilities being deployed on your matter are being applied effectively and whether their value to your case is worth the investment.
What Ethical Billing Actually Looks Like
Let me describe how this should work, because the correct model is not complicated — it is simply not universal.
When I use AI on a matter, my time directing the AI is billed at my rate, exactly as I would bill for time directing an associate. What the AI then does — the document review, the initial draft, the research — is performed in seconds and generates no additional billable time, because no additional time was spent. I then review the work product, edit it, and verify every citation, often by pulling each case directly on Westlaw and replacing the text with a linked citation. That process confirms the case is real, that I have read it, and that it accurately supports the proposition for which it is cited. The time I spend on that review is billed. If the total comes to three hours, the client is billed for three hours at my rate.
Compare that to what the same work would have cost without AI: twelve or more hours of associate time, plus my time reviewing the associate’s draft, plus my time verifying citations. The client’s bill is a fraction of what it would have been. The work product is not worse — it is better, because a senior lawyer’s judgment has been applied throughout rather than delegated to a junior associate for the substantive work and reserved only for the review.
That is the correct model. The culprit is the firm — or the individual lawyer — that looks at the three hours actually spent and fills the remaining twelve with time that was never worked. The AI completed what an associate would have billed fifteen hours to do. Billing as though the associate did the work anyway is not a billing judgment call. It is billing for time that did not exist.
Arizona’s Ethical Rule 1.5, like its counterpart in most jurisdictions, requires that fees be reasonable. The factors that determine reasonableness include, explicitly, the time and labor actually required to perform the work. When AI has reduced the time and labor required from fifteen hours to three, a fee built on fifteen hours is not reasonable — it is a fee built on a fiction. Bar associations are beginning to address this directly, and the lawyers who have not thought carefully about how AI affects their billing obligations will eventually be forced to by regulators who have.
The Questions Worth Asking
Most clients never have an explicit conversation with outside counsel about how AI affects their bill. Here is what to ask, at the outset of every significant engagement:
1. How is AI-assisted work billed? The correct answer is that the lawyer bills for time actually spent — directing and reviewing AI work — and does not bill for time the AI spent processing in seconds what an associate would have spent hours doing. If the answer is vague, or involves references to value billing without further explanation, ask the follow-up: are you billing for time actually spent, or for what the time would have been without AI?
2. What does your AI workflow actually look like? A lawyer who uses AI seriously should be able to walk through the specific steps: what the AI is directed to do, what comes back, how the work product is reviewed, and how citations and verifiable facts are confirmed before anything goes to the client. That workflow description tells you whether AI is being used responsibly — and whether the billing that follows it reflects actual time.
3. Do your invoices reflect AI use? A firm that has thought carefully about this should be able to tell you, on a given invoice, what was AI-assisted and what was not. That transparency is a signal that the billing has been thought through rather than carried over from a pre-AI model.
4. Is your total fee lower than it would have been without AI? This is the bottom line. If the answer is never — if AI use produces no change in what the client is billed — then either the AI is not being used on billable tasks, or the time savings are being filled with phantom hours. Neither is acceptable.
The Model That Actually Works for Clients
The billing approach that genuinely serves clients in an AI-enabled practice starts with a frank conversation at the outset of every engagement about what AI will be used for, how the time will be tracked, and what the client should expect to see on the invoice. It involves a genuine commitment to billing only for time actually spent — not for the time a task used to take before AI changed the math.
That commitment has to be explicit, because the default is otherwise. A firm that has made a serious investment in AI tools has a financial interest in recouping that investment. The firms that pass the savings to clients are making a deliberate choice to compete on value and on the integrity of their billing, rather than on the assumption that clients will not notice the difference.
AI investments should be designed to deliver better preparation and better results, and where they compress the time spent on tasks that would otherwise bill by the hour, that efficiency belongs to the client. That means some engagements cost less than they would have cost five years ago for equivalent work. It also means the work is better. Those two things are not in tension; they are the point.
If you have questions about how we approach AI and billing at Silver Cain — or about what you should be asking your current outside counsel — we are glad to have that conversation.
A Word About SilverCain
Silver Cain PLC was founded on the premise that businesses deserve both exceptional litigation experience and direct partner access — and that you should not have to choose between them. Leon Silver and Rebecca Cain have spent decades handling the most complex business and real estate disputes in Arizona and nationally. If you are evaluating counsel in Phoenix, we welcome the conversation.
