The $500-an-Hour Problem: Why AI Is Killing the Billable Hour

The founding partner of a 200 employee law firm says the real AI revolution in law isn't about replacing attorneys. It's about blowing up how they get paid.

Bob Kopka bills in six-minute increments. One-tenth of an hour. He often jokes about the absurdity of the billing system he’s worked in for three decades. "I’ll tell people, 'I saw you on the street and I said hello, so I sent you a bill for one-tenth of an hour,'" he says, laughing. While it’s a punchline, he uses it to illustrate a serious point: the traditional hourly model is fundamentally disconnected from actual value.

But here's the thing that keeps Kopka up at night: the AI tools his firm just adopted can compress a ten-hour research task into four hours. Under the old math, that's $3,000 in lost revenue per task. Multiply that across 100 attorneys handling thousands of cases per year and you're staring at a problem that no amount of six-minute increments can solve.

Kopka runs a defense litigation firm with roughly 100 lawyers spread across five states and about 200 total employees. His clients are insurance companies and self-insured corporations, the kind of organizations that send him a steady stream of slip-and-fall cases, auto accidents, and construction injuries. He's been at this for 30 years. And right now, sitting in Arizona where driverless Waymos cruise past his window, he's convinced that the legal profession is heading for its biggest business model shakeup since the billable hour became standard practice in the 1970s.

"Hourly billing first of all has no relation whatsoever to value," he says. "None whatsoever. And secondly, it's the enemy of efficiency."

The Efficiency Trap

The legal industry has an AI paradox, and it goes like this: every firm wants the productivity gains, but nobody wants to bill less for them.

A 2025 BigHand survey found that 100% of law firms reported AI and technology are already changing their pricing strategies. Every single one. Yet only about a third had actually adopted new pricing models to reflect those efficiencies. The gap between knowing the change is coming and doing something about it is wide enough to drive a Waymo through.

The numbers tell the story. Within any given practice area, 30 to 60 percent of billable hours go toward repeatable, rules-based tasks: summarizing medical records, drafting standard motions, searching for relevant case law. That's exactly the work AI handles best. And the efficiency gains aren't marginal. They're 50 to 90 percent reductions in time, depending on the task.

For Kopka, the math is personal. His firm licenses Lexis+ with Protégé, an AI-powered legal research platform from LexisNexis that can find relevant case law, suggest affirmative defenses, and summarize mountains of medical records in a fraction of the time a paralegal would need. They also use Filevine, a case management platform whose AI can help a lawyer set deposition goals, then track in real time whether those goals have been met during questioning.

"What the lawyers are telling me is that it makes them better lawyers," Kopka says. "And I've experienced that myself."

The catch? Better and faster lawyers who bill by the hour make less money. "To pay an additional charge to have attorneys be more efficient and then bill less doesn't make any sense," he says.

Here's a look at how AI is compressing the work without compressing the bills:

Task	Traditional Time	With AI Tools	Time Saved
Legal research for similar case precedent	8-12 hours	2-4 hours	60-75%
Summarizing medical records	4-6 hours	30-60 minutes	85-90%
Drafting initial motion or answer	3-5 hours	1-2 hours	50-70%
Deposition preparation	6-10 hours	2-4 hours	55-70%

Estimates based on industry reports and practitioner interviews

Subscription Models and the Death of 0.1

Kopka’s solution is blunt: stop billing by the hour entirely. He is pushing his insurance company clients toward what he calls the "Partnership Model," which utilizes a subscription-based set fee per file per month. Under this system, fees are tiered: more complicated cases carry a higher monthly fee, while less complex matters carry a lower rate.

The logic is simple. Around 95 to 96 percent of civil lawsuits settle before trial, according to Department of Justice data. Most of the others are won before a jury is ever seated. So the real question for a defense firm isn't how many hours you can bill. It's how fast you can get the case closed.

"The Partnership Model aligns the attorneys with the clients in the common goal of early favorable case resolution," Kopka explains. "It also provides an incentive to the client to move expeditiously in evaluating and settling each litigated case, since it saves them legal fees to do so."

This isn't just theory. His firm already tracks what he calls "cycle time," the average duration from opening a case file to closing it. When you're handling a portfolio of work for State Farm or CNA or Allstate, that metric matters more than any individual invoice line item. And AI, he argues, should shrink that cycle time. Which means both sides win: the client pays less overall, and the firm handles more cases in the same timeframe.

"We will be measured by the value of the outcome as opposed to the hours put into it."

He's not alone in this thinking. A Best Law Firms survey found that 72% of U.S. law firms now offer some form of alternative fee arrangement. Among firms with more than 50 lawyers, that number climbs to 90%. Flat fees are the most popular option, used by 73% of firms that have moved away from pure hourly billing. And the pressure from the client side is mounting: "AI discounts" are becoming a standard feature in legal RFPs, especially as corporations prepare for 2026 panel reviews.

Kopka takes this seriously enough that he just finished a planning call for a panel at the CLM Annual Conference, held this month at Disney's Coronado Springs Resort in Florida. The panel's topic? Changing billing models in response to artificial intelligence. Three insurance company executives and another law firm leader will join him. The conference draws about 4,000 attendees.

"Every one of the insurance company clients, they understand and they're adapting to it," he says.

The Guardrails Problem

If there's one thing Kopka wants other firm owners to understand, it's that adopting AI without guardrails is worse than not adopting it at all.

His firm put a blanket prohibition on using any AI tool that hasn't been vetted internally. No ChatGPT. No Gemini. No Claude. Nothing that sends client data into a public large language model. The only approved tools are the ones built into platforms the firm already licenses, where the data stays encrypted and private.

He has good reason to be cautious. The legal profession's AI horror stories are piling up fast. Since the Mata v. Avianca case in 2023, when a New York attorney cited six completely fabricated cases generated by ChatGPT, the problem has exploded. Researcher Damien Charlotin now tracks over 600 instances of AI hallucinations in court filings nationwide. In 2025 alone, the pace accelerated from two cases per week to two or three per day. A Stanford RegLab analysis found that some AI models hallucinate in one out of every three legal research queries.

"Lawyers using them have gotten into trouble because they're not accurate," Kopka says. "Lawyers have cited cases that they've got from ChatGPT that aren't really there, that are not published cases, and then they get sanctioned for that."

The sanctions have gotten severe. In September 2025, a California court fined an attorney $10,000 after finding that 21 of 23 case quotations in his appellate brief were fabricated by AI. A Denver attorney accepted a 90-day suspension for the same kind of mistake after texting a paralegal that, "like an idiot," he hadn't checked ChatGPT's work.

Here's a snapshot of the growing fallout:

Case	Year	What Happened	Consequence
Mata v. Avianca (New York)	2023	6 fabricated cases cited from ChatGPT	Sanctions against attorney
Arizona Social Security case	2024	12 of 19 cited cases were fabricated	Sanctions by federal judge
Morgan & Morgan / Goody Law (Florida)	2025	Internal AI platform hallucinated case citations	$5,000 in combined sanctions, attorney removed from case
Noland v. Land of the Free (California)	2025	21 of 23 quotations in appellate brief fabricated	$10,000 fine, state bar referral
Denver attorney suspension	2025	Filed motion with unchecked ChatGPT citations	90-day license suspension

Sources: ABA Journal, Cronkite News, CalMatters

The vetted platforms Kopka's firm uses work differently. LexisNexis Protégé grounds its AI responses in actual published case law, includes Shepard's citation verification, and runs on models that are exclusive to LexisNexis customers. Filevine's AI operates within its own secure, HIPAA-compliant, SOC 2 Type II certified environment. Neither tool feeds client data into a public model.

"These are tools that are vetted by the firm. They are secure. They're not public," Kopka says. "The large language models don't incorporate the work in a public sphere in order to help them grow."

The Human Ceiling

For anyone wondering whether AI will replace lawyers entirely, Kopka isn't worried. Not because the technology can't get smarter, but because the systems surrounding legal work won't allow it.

He recently presented to an insurance company about anticipated AI regulation in the industry. The message from state insurance commissioners was clear: coverage decisions, valuation decisions, bad faith determinations, all of it has to involve a human being. The National Association of Insurance Commissioners has been coordinating state-level guidelines that prohibit using AI as the sole decision-maker in claims processing. Multiple states introduced bills in 2025 requiring human review of any algorithm-driven coverage determination.

"While AI can provide data, information, options, decisions need to be made by a human being," Kopka says. Courts, arbitration panels, mediations: all of them require human representation. And he doesn't see that changing.

"I don't see an artificial intelligence being able to take over that job," he says. "At least for the foreseeable future."

What he does see is a different kind of attorney. Less paper-pusher, more strategist. One who spends less time grinding through records and more time actually thinking about the case. "When you're doing research and you're looking for the very case that may just be the one case that's very similar to the one that you have now," he says, the AI "allows us to find that case much quicker." Where a lawyer used to read case after case after case hunting for the right precedent, Protégé can narrow the search and surface it in minutes.

The Protégé system also helps with the kind of creative legal thinking that separates good lawyers from great ones. Kopka describes asking the AI about potential affirmative defenses for a case and getting suggestions he might not have considered on his own, or at least wouldn't have found as quickly. It's not replacing judgment. It's expanding the menu of options a lawyer can consider before exercising that judgment.

The same goes for support staff, though the picture there is more complicated. Kopka remembers when everything he wrote had to be retyped by a secretary on a typewriter. Mistakes meant white-out or starting over. Now he dictates into his phone and gets a clean transcript. Electronic filing has eliminated the clerks who used to physically carry documents to the courthouse and bring back stamped copies.

"You've seen the efficiencies already in the last 30 years," he says. "And AI is just going to enhance those efficiencies."

Will that mean fewer support roles? Yes. AI adoption among legal professionals more than doubled between 2025 and 2026, jumping from 31% to 69%, and paralegals handling document summarization are already feeling the squeeze. But Kopka resists putting a specific number on it. "I don't know about 80%," he says, responding to a recent Anthropic report suggesting AI could affect 88% of legal support roles. "I'm not going to give it a number. But clearly, lawyers will need fewer support staff in order to get work out. Still need somebody."

Accept It, Adapt to It, Adopt It

Kopka's practical advice to anyone starting a law firm or considering how to integrate AI comes down to six words: "Accept it, adapt to it, adopt it."

He doesn't sugarcoat the disruption. But he doesn't think you get to opt out of it, either.

He compares it to the driverless cars he's been watching pass outside his window in Arizona. "People that say I'm never going to get into a car that doesn't have a driver, they are going to get into a car that doesn't have a driver," he says. "It just depends on when. It's not if anymore."

The cost is real. Protégé wasn't cheap. Filevine's AI features added to the bill. But Kopka sees these as investments that only make sense if the business model changes to match. A firm that pays for AI tools and then keeps billing by the hour is essentially subsidizing its clients' savings while eating the expense itself.

"We have to be able to monetize it," he says. "We have to be able to say there's a return on this investment."

His trepidation, and he admits he has some, isn't about the technology. It's about the ripple effects. "I don't know how it's going to affect overall employment," he says. "When you think about programmers, certain jobs will go away. They'll be done by computers." He pauses. "But I think ultimately computers can't make the kind of decisions that human beings make. And so, decision makers will always have to be human beings."

Even his doctor has gotten into the act. On a recent visit, the physician asked if he could have AI record their conversation. The recording became the visit notes, saving the doctor time after each appointment. "Those are his notes," Kopka says, marveling a little at how the same wave is hitting every profession at once.

There's a version of the future where this all works out neatly. AI lowers the cost of legal services. More people who couldn't afford an attorney suddenly can. Firms compete on outcomes rather than hours, and the incentives finally align between lawyer and client. Cycle times shrink. Disputes resolve faster. The old jokes about lawyers billing for handshakes start to sound like ancient history.

"I think AI eventually will lower the cost of legal services and will provide incentives and in a better way to measure the value that clients are getting from law firms. Not billable hours anymore, but outcomes. Positive outcomes."

Kopka sounds genuinely optimistic about that version. He's building his panel presentation around it. He's been on the phone with insurance executives who agree. The clients want it. The math supports it. The technology already exists.

The only question, really, is who moves first. And how many six-minute increments the rest of the profession will burn through before they figure out that the clock on the wall isn't the one that matters anymore.

To learn more about Bob Kopka and his law firm, you can visit his website here: Kopka Law

Nothing in this article constitutes legal advice. The views expressed by the individuals quoted are their own and are provided for informational and educational purposes only. No attorney-client relationship is created by reading this content. If you need legal guidance, consult a licensed attorney in your jurisdiction.