Get your firm's free AI report card
BLOG POST · SOLO & SMALL FIRM PRACTICE

If AI Does It in Minutes, What Do You Charge?

TL;DR
  • Clients pay for your judgment, your sign-off and your liability, so charge a fair, flat fee rather than billing the minutes.
  • Price transactional work on deal size, risk and the cost of a missed clause, and write the scope down.
  • ABA Formal Opinion 512 warns a flat fee isn't automatically reasonable if AI makes the work much faster, so check your rules.

AI can produce the first draft of a contract, a demand letter or a research memo in minutes. If you bill by the hour, those minutes are now most of your invoice. So what do you charge for work a machine started and you finished? Here's how to think about it, using a pricing course from the Solo Practice University archive that predates large language models.

The signature is the product

Start with what the client actually gets when a lawyer delivers AI-assisted work. It isn't keystrokes. It's a professional who read the draft, decided what was wrong with it, fixed it, and put their name on it.

John O'Connor, founder of Associate and SPU's Professor of AI Fluency, put it this way:

"The judgment of the real person, and the fact that they are signing off on the work, carries weight beyond just time dedicated to that one matter."

O'Connor borrows a Japanese idea to make the point. A hanko is the personal seal used in Japan to sign and authorize documents.

"Attorney work of any kind carries what the Japanese would call a hanko, a signature of authenticity."

The seal is what makes the document count. The same is true of a lawyer's sign-off, and it comes with something AI doesn't carry: liability.

"If the attorney is liable for the work, which it's crucial to remember they are, they needn't be shy about charging a fair, flat fee for it."

That's the thesis of this post. The rest is how to set that fee without falling back on the clock.

What the client is buying

The lawyer's AIA process built on your expertise, not a generic chatbot
JudgmentWhat matters, what to ask, what to leave out
Sign-offYour name and your liability on the result
A fair, flat feePriced on what's delivered, not the minutes it took
The formula is John O'Connor's: "They hire that lawyer's AI, plus judgment, plus signoff." A hanko is the seal used in Japan to sign and authorize documents.

One option firms are discussing: bill the agent's time

Not everyone thinks the answer is to drop hourly billing. Seth Fenster, who writes about AI and law, made the case for keeping the hour and splitting it in a LinkedIn post titled "Law Firms Should Bill by the Hour for AI Agents". Fenster starts with the problem every hourly lawyer using AI has run into:

"If I spend 10 minutes on your documents and Claude spends the next 50 finding issues and drafting, I can only bill my 10."

Fenster's answer: "BILL THE AGENT'S TIME SEPARATELY," with a hypothetical of "$1,000/hr for the lawyer and $100/hr for the firm's agent." In Fenster's sample invoice, 10 lawyer minutes ($166.67) plus 50 agent minutes ($83.33) comes to $250, against $1,000 if the lawyer had worked the full hour. Fenster labels that "75% client savings."

Fenster's hypothetical: one hour of document review

Lawyer works all 60 minutes$1,000
Lawyer 10 min + agent 50 min$250

Lawyer 10 min at $1,000/hr = $166.67   Agent 50 min at $100/hr = $83.33

Rates and invoice are Seth Fenster's hypothetical, from his LinkedIn post. An illustration of the model, not a recommendation.

The agent in this model isn't a generic chatbot: "It's not out-of-the box vanilla Claude. It's an agent running on the firm's harness, skills, and legal knowledge." Fenster, who says they have opposed the billable hour, concludes: "when the agent is the one putting in the hours, it's actually a pretty good model!"

The SPU course behind this post is from the opposite camp. It argued that lawyers sell things, not time. By that logic, there's no reason to meter anyone's hours, human or otherwise. Which fits depends on your practice, your clients and your jurisdiction's fee rules.

Don't reverse engineer an hourly rate

The course is Fixing Your Fees, Fixing Your Practice, taught for SPU by Jay Shepherd, whose firm had dropped hourly billing at the time. (We covered the case against the hourly model in AI makes you faster, the billable hour makes that a pay cut, including why rates are not prices: a rate tells the client nothing until someone guesses the hours.)

The most common mistake the instructor saw from lawyers trying fixed fees was doing the hourly math anyway, after the fact:

"So don't go back and say, 'Oh, gee, it took me 16.2 hours.' And my normal hourly rate is x and 16.2 times x is what I would have made. And I only charged this much of my fixed fee. I lost money. This stinks. … You can't do that."

Time and value are separate things. In the lecture's words, "You have to unlink the concept of the time spent with the value that the client puts on the work you're doing."

The course applied this in both directions. If a matter priced at $9,000 took more work than expected, the price didn't move:

"So what? It doesn't change your costs. It does not change the value that the client is receiving for this work. So why should it change your price?"

AI runs the same logic in reverse. If the work takes less time, the value to the client hasn't dropped either. Price from hours and every better tool cuts your fee. Price from value and you keep the gain. (For how the course lands on a number, see the 14 steps for setting a fixed price.)

Transactional work: price the stakes, not the paperwork

Transactional work is where AI saves the most time, so it's where hourly billing hurts the most. The instructor thought it should also be the easiest to price: if litigation, the hardest thing to price, can be priced, everything else can too.

The lecture's examples were real estate and wills. A $2 million house in the suburbs might not take much more work than a condo in the city, so why charge the same for both? On wills, the instructor took aim at lawyers who quote a flat price for a simple will, then switch to hourly when the estate gets complicated. The hypothetical: a $700 simple will next to "Mr. Moneybags," who wants a complex trust for a $3 million estate.

"Doesn't Mr. Moneybags deserve the certainty?"

The verdict: "don't use hourly billing as a cop-out." Offering hourly as the alternative tells the client you don't know what your work is worth.

For AI-assisted drafting, the draft may be fast. The deal size, the risk and the consequences of a missed clause are not. Price those, and write the scope down so the price holds (our post on fixed-price agreements and change orders covers how the course handled that).

Advice work: the hardest thing to price

The instructor was candid that advice work gave the firm more trouble than litigation:

"Advice is hard. You know, you get a phone call with a question about something and it's oftentimes very difficult to tell if at the beginning, if this is going to be, you know, two phone calls or six months of involved hand holding and advising."

The firm's answer had two parts. Regular clients got a monthly membership: unlimited advice for a fixed price, with higher tiers that added drafting and training. (The SPU archive covers this model in monthly retainers and the solo practitioner.) Occasional clients got a limited membership: advice on one issue for 30 days, at a price set once the firm understood the issue. That fix came after months of what the lecture called "Groundhog Day style phone calls" for a single fee.

"If you want my advice for a few days, a couple of weeks, that's one thing. If you want my advice for half a year, well, you should be willing to pay more for that."

The firm kept advice prices on the lower side, because clients were more price sensitive about advice than about being sued. The rule: "when people are less price sensitive, charge more, when they're more price sensitive, charge less, and you will make more."

AI changes the inputs here, not the structure. The client is still buying access to your judgment over a period of time, and a time-boxed or monthly price fits that better than a meter.

Menus and the commodity trap

Cheap AI output tempts firms to post a price list and compete on price. The course pushed back on both. On menus:

"There's no reason why you need to have a menu of prices like on your website or your windows. You know, these aren't happy meals, folks."

The right price for one client isn't the right price for another, so you can't quote it before you know the matter. On commodity pricing, the course drew a line that matters more now than it did then:

"If something is routine and not that important and not that urgent and that any trained chimpanzee could do, you shouldn't charge a premium for that."

But the instructor rejected the idea that legal practice itself is a commodity. Sugar and crude oil are undifferentiated. Lawyering isn't: "There's nobody in the world… who does lawyering, whatever type of law that you practice, exactly like you do." The lecture also warned solos against discounting their way into a price war, because larger competitors can always afford to undercut you.

The AI-era version: sort your work honestly. Some tasks really are routine, and AI will push their price down whether you like it or not. Others depend on your judgment, your niche and your sign-off. Charge a premium for those, and make the difference visible to clients, because when clients don't perceive value, they shop on price.

What the ethics opinions say

ABA Formal Opinion 512 (July 2024) says lawyers who bill hourly must bill for their actual time, not the time the tool saved. It also says a flat fee isn't automatically reasonable: if an AI tool lets a lawyer finish a task much faster, charging the same flat fee as before may be unreasonable under Rule 1.5.

So "flat fee" isn't a way around the reasonableness rule. Check your own jurisdiction's rules before you change how you charge.

What clients are hiring

The course kept coming back to what the client is buying. It isn't the hours, and increasingly it isn't the first draft either. O'Connor's formula:

"They hire that lawyer's AI, plus judgment, plus signoff."

Price that, and the minutes stop mattering. For more on how these shifts are changing small-firm economics, see our demographic and economic trends coverage. Whatever you decide, make sure your website describes how you charge the way you actually charge. SPU offers a free law firm website review.

Take the full course

The complete Fixing Your Fees, Fixing Your Practice course is in the SPU community. It covers pricing litigation phase by phase, market positioning, risk-shifting fees, fixing pricing mistakes and the ethics of fixed fees, plus an exercise where you price one of your own services and get feedback from other members. SPU membership gets you the community, monthly strategy sessions and the full course library.

THE WEEKLY SPU BRIEF

Get one practical lesson for your firm every week.

Free, from practicing lawyers. Every issue includes a lesson from the archive. Unsubscribe anytime.

ABOUT THE AUTHOR
The SPU Team

The SPU Team writes and edits Solo Practice University's articles. Susan Cartier Liebel founded SPU in 2009 to teach lawyers the practical side of building a practice, the part law school leaves out, with courses taught by practicing lawyers. Today SPU is a community for solo and small-firm lawyers, with an AI Fluency program for firms putting AI to work. Nothing here is legal advice: check your own jurisdiction's rules.

About Solo Practice University →