Featuring insights from Cuatrecasas partner Omar Puertas, this piece shows how generative AI is already elevating legal reasoning and client value. His message is optimistic and practical: teams that start experimenting now compound skills, speed, and a durable competitive edge.
When Omar Puertas first tried ChatGPT on 30 November 2022, he didn’t ask for a poem.
“I asked for an NDA subject to New York law,” the Cuatrecasas equity partner recalls. The bot delivered.
That single prompt pulled a tech-obsessed arbitration lawyer down a rabbit hole of machine-learning courses, late-night prompt engineering, and 30,000 internal test prompts with 100 colleagues.
Today Puertas straddles two worlds, board-level legal strategy and deep-dive AI experimentation, and he’s adamant: This technology is different.
What generative AI actually changes
Most legal tech tools help you do work faster. Generative AI helps you think better.
“You don’t use the technology, you reason with it,” Puertas says. “It’s a mode of intelligence that restructures legal reasoning itself.”
That means the big wins aren’t automating NDAs or first drafts (though those are real). They're asking the model to play devil’s advocate, surface counter-arguments, and expose hidden risks in clauses you already wrote.
The Innovator’s Dilemma is now in pinstripes
Clayton Christensen’s classic theory predicts incumbents ignore disruptive tech because early versions look weaker and less profitable. Puertas sees the pattern in law:
- The model hallucinates?
- It’s stochastic?
- You still have to review everything?
“Rationally, a firm focused on today’s profit margin will postpone investing,” he says. “By the time the tech is good enough, the train has left the station.”
Where AI excels and where it still struggles
Understanding exactly where AI currently adds value, and where caution is still required, is key to integrating it into legal workflows.
AI’s strengths:
- Creative reasoning and argument refinement
- Rapid summarisation of extensive records (Platforms like Emma excel at quickly distilling large data sets into actionable insights.)
- 'Destroy my clause' analysis that surfaces blind spots
Areas still needing improvement:
- Case-law retrieval (“It hallucinates citations,” Puertas notes.)
- Tasks dependent on live, proprietary databases
- Automated workflows lacking human oversight
Cuatrecasas formalised these insights into clear usage guidelines: green-lit scenarios for immediate adoption, caution zones that require extra scrutiny, and red-flag areas where AI isn't yet reliable.
From 90% accurate to 100% valuable
Puertas’ favourite thought experiment is this: a ChatGPT-generated NDA might only reach 90% of his own professional standard; but it costs mere pennies compared to his hourly fee.
“That 90% fundamentally reshapes the economics of legal services,” he argues. Clients will happily continue paying premium rates, but only for that critical 10% that delivers strategic, high-stakes value.
Why associates shouldn’t panic
A student once asked Puertas whether his six hours AI-prompting training could replace his 25 years of experience.
“I don’t know,” he laughed, “but the curiosity you’ve just demonstrated will make you a far better lawyer than I was at your age.”
A well-guided associate using AI can:
- Ask the model to break down complex indemnity clauses at their own comprehension level.
- Generate potential counter-arguments before facing a partner’s critique.
- Quickly benchmark their analysis against thousands of anonymised clauses.
Routine drudgery decreases, while the opportunity to rapidly develop analytical skills significantly increases.
The path to 2035 and why urgency matters today
Puertas believes the adage: we tend to overestimate technology's impact within one year but greatly underestimate its effect over a decade.
Within two years, he expects AI-enhanced, faster, cheaper, better work will become essential for any law firm targeting premium business. Five years from now? Possibly subscription-based legal services allowing clients to directly query a firm’s decades of accumulated expertise via a custom-trained AI.
What’s undeniably clear: firms that delay experimenting with AI risk finding themselves in a crowded and expensive race to catch up.
One final piece of advice
“If you’re not experimenting yet, you’re already behind,” Puertas warns.
Rethink your definition of 'technology' and start actively reasoning with AI today. Book a call today to explore how Emma Legal can streamline your legal due diligence and transform your approach to AI-powered legal analysis.
Other related podcasts
AI for lawyers is here. Now what?
AI for lawyers is here. Omar Puertas, partner at Cuatrecasas, explains where gen-AI adds value, where it fails, and why firms must start experimenting with Emma Legal.
Read
AI leverage is the new deal currency for PEs
Featuring insights from Connor O’Hearn, this article explores how AI is becoming the new currency of leverage in M&A. O’Hearn, founder of MAAI Services, outlines how lean deal teams are outpacing larger firms by using tools like Emma Legal to automate diligence, uncover insights, and compress timelines. He argues that AI isn't just about efficiency. It’s a career accelerant, a strategic edge, and a sign of maturity that LPs are starting to expect.
Read
Why Cybersecurity is a dealbreaker in M&A
Diving into the increasingly critical role of cybersecurity in M&A. From AI-driven due diligence to the evolution of virtual data rooms, we explore how technology, risk management, and security culture intersect in modern deal making.
Read