Counting ROI or Chasing Hype: Stephanie Curcio on the True Test of AI in Patents

In the most recent episode of IP Innovators, host Steve Brachmann sits down with Stephanie Curcio, CEO and co-founder of NLPatent, to unpack how AI is reshaping prosecution, search, and the overall workflow across patent professions. Curcio, who began her career in traditional patent drafting and prosecution, explains how early concept-based AI search tools convinced her the profession was on the verge of a seismic shift.

As firms now rush to adopt AI, Curcio argues that the real challenge is no longer discovering what AI can do but understanding what it actually delivers. Efficiency, accuracy, ROI, ethics, and the quiet return of the build-vs-buy debate form the core questions firms must answer—not the hype around AI features themselves.

The Efficiency Illusion—and Why the Patent Profession Must Measure, Not Assume

Patent professionals often take efficiency as a given benefit of AI. Curcio challenges that assumption head-on.

She cites a recent study about AI tools for software developers that revealed a surprising outcome: “These tools actually decreased efficiency across the subsection of professionals that they were studying,” she said. Even more striking, the study found that users “thought they were being more efficient, but the study revealed that they were actually less efficient.”

For patent teams, her message is clear: Never confuse the feeling of speed with actual gains.

That’s not to say AI tools won’t bring efficiency, however. To Curcio, the secret lies in implementation. “Introducing certain efficiencies in one aspect might introduce inefficiencies in others,” Curcio reiterates. “So I think it’s a matter of checks and balances and ensuring that we’re deploying in the right way that’s going to prove to be effective across the workflow of the end user and not just create more problems.”

Curcio urges firms to measure ROI with rigor—time per task, error reduction, improved claim clarity—not gut instinct. “Efficiency is only one aspect of the equation…the accuracy of your end work product and the quality of that work product can certainly be quite impacted by AI solutions.”

In other words, a tool that marginally slows a drafter but produces a stronger, more defensible patent may offer far higher value than a tool that produces faster, but weaker, output.

Responsible Adoption Requires Balance, Not Extremes

Curcio sees firms falling into two equally risky categories: those that over-police AI adoption, and those that barely evaluate it at all.

At one end of the spectrum: “Here’s my security questionnaire…we need to see all of your penetration tests and your SOC2 reports, and we need to put you through the ringer.”

At the other: “We just need to move on this. So let’s sign where we need to sign and let’s go.”

Both approaches fail. One stalls innovation; the other invites avoidable risk. For Curcio, the winners will be firms that balance the extremes in a way that understands the technology they’re adopting.

This is especially urgent, she notes, because even legacy tools now ship AI features quietly. “All of a sudden [they turn] around and, ‘Ta-da! We’re an AI company.’” And with those surprise features come new, often unexamined terms of service. “What does that mean for the consumer and how is their data being used?”

The Quiet Return of Build vs. Buy—And Why It Could Reshape Competitive Advantage

One of Curcio’s most compelling observations is her claim that the build vs. buy debate is back, despite largely disappearing years ago.

In the early 2010s, some firms experimented with building internal tools, but the cost of engineers, maintenance, and infrastructure made long-term investment impractical. Curcio describes it bluntly: “The whole architecture was too expensive…it’s not the business of law.”

But in 2025, AI, low-code tools, and off-the-shelf models have entirely changed the economics.

“This build vs. buy question has come up again, but nobody’s really talking about it.” She sees firms quietly building internal, hyper-specific tools that vendors would never create—capitalizing on the new reality that the barrier to entry “has decreased significantly.”

This gives larger firms—with resources, data, and internal IT talent—a profound structural advantage over smaller competitors.

The takeaway is pivotal: The next competitive edge in patent practice won’t come from who adopts AI fastest—but from who adopts it most intelligently. Firms that treat AI as a monolithic off-the-shelf solution risk becoming indistinguishable from their peers. Firms that adopt strategic, targeted, custom tools—whether built in-house or from a carefully-selected partner—may develop capabilities competitors cannot easily replicate.

What Patent Professionals Should Be Preparing For Next

Curcio believes the future of patent technology will favor modular ecosystems—not all-in-one platforms.

She describes a coming landscape where platforms “connect the best solutions together” so firms can assemble customized stacks that fit their workflows exactly. In a profession with wildly different needs between biotech, software, hardware, litigation, and portfolio strategy, this modularity will matter.

“You can’t boil the ocean,” she says. “But if you focus on one problem, you have the possibility to be the absolute best at solving that problem.”

This hints at a future where:

• Firms shift toward modular tools that fit into their workflow
• Internal tools fill gaps between vendor capabilities
• Integration becomes a core strategic competency
• Data flow and security become board-level issues
• ROI measurement becomes as essential as billing hygiene

And most importantly, the firms that embrace measured, intelligent, ROI-driven experimentation will lead the next era of patent practice.

Why This Moment Matters More Than Ever

Curcio’s story is, at its core, about perspective. She entered AI for IP before GenAI existed—when AI search models were still quietly outperforming human experts behind the scenes. Now, she watches the landscape shift daily as both vendors and law firms sprint toward new capabilities.

But for all the excitement, she warns that many practitioners still misunderstand AI entirely. “It never ceases to amaze me how little people know about AI solutions…what is AI and what is not AI.”

That knowledge gap, she suggests, is where the biggest risks—and the biggest opportunities—now lie.

For firms willing to deepen their understanding, measure their ROI honestly, and approach AI with both optimism and discipline, the next decade of patent practice could be foundationally defined by higher productivity and innovation.

IP Innovators is proudly sponsored by DeepIP—the patent intelligence platform for in-house teams and external counsel. Learn more at deepip.ai.