As a law student at Duquesne University, Von Wooding knew technology was coming to the legal profession, and he wanted to be at the bleeding edge.
"My first year of law school, I went and worked in Silicon Valley for a bitcoin brokerage, which was completely unheard of at Duquesne,” he said. "Most of my classmates were probably thinking I'm crazy.”
A year after graduating, he launched Counsel Stack, which leverages artificial intelligence to help lawyers do research and form arguments.
The startup raised $200,000 from a small new investment firm, Brown and White Ventures, based in Tampa, Fla., which was seeking to capitalize on AI solutions. Small practices in Pittsburgh and Morgantown became early adopters. The software will soon launch publicly for $300 a month.
But experts at Mr. Wooding’s alma mater warned of risks, including bias and hallucinations — the tendency for AI to conjure false information out of thin air.
“AI has not proven itself,” said John Slattery, executive director of Duquesne’s Carl G. Grefenstette Center for Ethics in Science, Technology, and Law. “There is a lot of potential here, but I think there's a lot of hesitancy from the top brass in the legal profession.”
Lawyers who use the software said it has saved time and money, which ultimately brings down costs for their clients. But they rarely tell those customers, or the courts, that AI was used — an omission that runs counter to guidance from the Pennsylvania Bar Association and some judges’ policies.
If people view the law as essentially one big database, then artificial intelligence may be just the tool to parse it, Mr. Slattery said. But legal work is also a world of specificity and meticulousness, where words have precise meanings that AI might not understand, he said.
From hesitant to optimistic
Even lawyers with good intentions could introduce bias into their work without carefully understanding, vetting and fact-checking the work of AI, Mr. Slattery said.
“It's a bit of a minefield to go into right now.”
Law dean April Barton was a bit more optimistic.
“It's very exciting to see that our graduates are at the forefront of innovation in this space,” she said. “In the case of Von, he's helping to shape the future of the profession, and he's doing it, I think, in a very ethical and responsible way.”
Ms. Barton acknowledged she doesn’t have direct insight into how Counsel Stack operates, but said the school tries to emphasize the ethical use of AI. Duquesne has launched a handful of AI-related initiatives, including a law and computing concentration that offers computer programming courses, and more specific classes on prompt engineering.
“We're all kind of grappling with this in real time and figuring it out,” Ms. Barton said.
One of Pittsburgh’s oldest intellectual property firms, Webb Law, which has been protecting inventions since 1845, is also experimenting with artificial intelligence.
Not everyone there is sold.
“Our paralegals, our attorneys, are looking at it and trying to verify what's actually accurate and factually can be used in the first place,” partner Cecilia R. Dickson told the Post-Gazette late last year.
Another Pittsburgh firm, Buchanan Ingersoll & Rooney, built its own AI system using a variety of open source software. Jeff Lagana, the firm's director of enterprise operations and security, said they built in safeguards to ensure confidentiality and reliability.
But he acknowledged: “From a legal standpoint, you're going to potentially lose authenticity if you just rely on AI to do everything for you.”
‘Just the way things are going’
A recent Carnegie Mellon University dropout took a more disruptive approach.
Before his 18th birthday, Ethan Hilton had positioned his own legal tech startup Caseflood.ai to take in thousands of calls for law firms around the country, replacing much of the administrative work previously done by human employees. He isn’t shy about that impact.
“If you’re an attorney in a high-volume practice area and you want to replace your entire admin staff, go to Caseflood.ai — we’ll help you out,” Mr. Hilton wrote on LinkedIn in January.
He told the Post-Gazette the post prompted some backlash. “I've gotten some flak for that,” he said.
But the young founder also described the work he’s automating as “not extremely high skilled labor” that is “ripe for replacement with automation.”
“The value proposition is clear to business owners, right? The calculus is easy: You can spend $400,000 to cut a $40 million a year division, and that's kind of just the way that things are going.”
After spending less than four months at CMU’s Tepper School of Business, Mr. Hilton said he was already solving real world problems and had earned a spot in the Silicon Valley accelerator Y Combinator. “I can always go back to school,” Mr. Hilton said.
Virtual legal assistants
The legal tech world has been chilled by a handful of high profile embarrassments, mostly involving lawyers who cited false cases generated by ChatGPT. At Counsel Stack, Mr. Wooding said he limited the software to avoid similar bungles.
“Lawyers, as subject-matter experts, always remain in the driver's seat,” he wrote in a company announcement.
Dayton Meadows, who began using Counsel Stack at his practice in Morgantown, said the platform replaced his existing management software and became “an invaluable tool” for research. Rather than replacing human work, he said it empowered his small team, which includes just one other attorney, a paralegal, and an intern.
“It makes you exponentially better,” he said.
Lawyers who use the system — billed as a set of “virtual legal assistants” — said they aren’t necessarily required to share with clients that they’re using an automated product.
“I'm not aware of any disclosures for having to state the use of an automated platform,” Mr. Meadows said. “But as an attorney, you still have your same standard of care. You still have your same obligations to ensure that that case research is correct.
Even if it takes some time to fact-check AI’s outputs, he said the system is saving time on the front end, which ultimately provides a cheaper delivery for his clients. He also draws a distinction in the cautionary tales of lawyers who have misused AI.
“They were not punished for using AI,” Mr. Meadows said. “They were punished for providing inaccurate law — so using the tool incorrectly.”
Jim Campese, a fellow Duquesne law alum who runs his own practice in Pittsburgh, takes a similar stance.
“People have been lazy since the dawn of time,” he said. “You have to qualify everything that leaves your desk and any work product that comes out of your firm. You're responsible for it the same way you would be if you delegated that work product to another person.”
When he first started practicing, Mr. Campese said there were lots of “goose chases” where he’d spend hours researching something that turned out to be useless. AI helps him arrive at that verdict faster.
“I can give it an entire case file and say ‘What patterns do you notice? Summarize it for me,’ and it can do that. The alternative is that I spend half my day doing it, and I might not find anything.”
Mr. Campese said the collegiate ethos he and Mr. Wooding shared at Duquesne drove him to want to use technology to help clients.
“It's been great to see Von develop, and that feels weird to say, because I'm only a couple years older than him,” he said.
But like Mr. Meadows, Mr. Campese said he doesn’t feel compelled to tell clients he’s using AI.
“Back when this technology was emerging, that would have been something that you would have wanted to include, because it would have been more controversial,” Mr. Campese said.
“Nobody wants to spend a million dollars on their lawyers, and if you can avoid that … then you're going to be doing right by your clients, and they're not going to mind.”
Legal concerns
The rules around the use of AI in courtrooms remains in flux.
An Ohio judge flatly banned the use of AI in preparing legal filings, while a Hawaii judge is requiring disclosure. Others are requiring proof that lawyers fact-checked AI outputs to ensure accuracy.
Those efforts are butting up against a seemingly inevitable march of progress, as primary database tools already used by lawyers, LexisNexis and Westlaw, have introduced AI features of their own.
Even the Ohio ruling allows a carve out for these tools: “The court does not intend this AI ban to apply to information gathered from legal search engines, such as Westlaw or LexisNexis, Internet search engines, such as Google or Bing.”
Pennsylvania’s first rulings on AI did not include exceptions. Lawyers “must, in a clear and plain factual statement, disclose that AI has been used,” one judge required in 2023. A year later, another judge issued the same requirement specifying the use of “generative” artificial intelligence.
The Pennsylvania Bar Association issued extensive guidance on the use of AI in 2024, warning about privacy, accuracy, and even billing.
On transparency, the recommendations were clear:
“Lawyers must communicate with clients about using AI technologies in their practices, providing clear and transparent explanations of how such tools are employed and their potential impact on case outcomes. If necessary, they should obtain client consent before using certain AI tools,” the guidance states. “Lawyers should be transparent with clients, colleagues, and the courts about the use of AI tools in legal practice, including disclosing any limitations or uncertainties associated with AI-generated content.”
Mr. Wooding, who said AI helped him pass the bar exam in Washington, D.C., acknowledges some of the tool’s imperfections.
“There are a lot of questions about, what is AI? Is it reliable? What if it makes up a citation and I get in trouble from the judge?
“Those things are all top of mind, and they're all very valid concerns,” he said.
First Published: February 15, 2025, 10:30 a.m.
Updated: February 17, 2025, 2:08 p.m.
