AI Won't Capture the Value. Practitioners Will.
Why the economics of AI in patent practice come down to expertise, not efficiency.
Welcome back to Behind the Claims.
The question everyone is asking about AI in patent practice is economic: who gets the savings? Is it the clients who want lower fees, or firms that invested in the tools?
It’s the right question. But the panel we convened for our recent IPWatchdog webinar kept arriving at a different one: what are you actually trying to accomplish with this technology—and do your tools reflect that?
Across an hour of conversation with DeepIP’s Co-Founder & CEO François-Xavier Leduc, joined by Ryan Phelan, Eli Mazour, Bryan McWhorter, and Gene Quinn, a clearer picture emerged. The practitioners capturing the most value from AI aren’t the ones moving fastest—they’re the ones who stayed in control of the legal judgment while letting AI do everything else. The bottlenecks shifted. The expertise didn’t.
Below, we break down what the panel said, what the data showed, and what it means for how patent professionals should be thinking about their workflows right now.
This Week in IP
The economics of AI in legal practice extend well beyond patent work. These three pieces capture where the broader profession stands, and where the pressure is building.
In-House Counsel Are Already Pushing for New Pricing Models
The 2025 ACC/Everlaw GenAI Strategic Value for Corporate Law Departments report, drawing on 657 in-house legal professionals across 30 countries, found that nearly 60% of respondents reported no noticeable savings yet from their outside counsel using GenAI on their matters—with 58% pointing to a deeper issue: law firms haven't adjusted pricing to reflect GenAI-driven efficiencies. At the same time, 61% said they plan to push for change in how legal services are delivered and priced from firms that use GenAI.
The transparency gap between what AI is delivering inside firms and what clients are seeing on invoices is where the pressure is building—and it maps directly to the conversation our panel had last week.AI Is Saving Time. The Billable Hour Is Catching Up.
Bloomberg Law's recent reporting on Texas law firms captures the live tension playing out across the profession: there is little debate that generative AI is saving law firms time. The question now becomes: who will save money? The piece documents firms treating AI costs as overhead, others passing savings directly to clients, and others still figuring it out.
Firms will start to offer different tiers of human review of AI tools, and what's key is they make such tiers understandable to clients—flagging possible risks and shortcomings. Sound familiar? It's almost word for word what Ryan Phelan described in our session.The Billable Hour's Collision Course with Client Expectations
The Thomson Reuters Institute's Law Firm Rates Report 2026 surfaces a counterintuitive finding: while firms have rushed to invest in AI to justify higher rates, regardless of whether firms discount aggressively or hold firm on realization, they're collecting roughly the same amount per hour—leaving firm leaders to grapple with where competitive advantage actually lives.
If efficiency gains aren't flowing to margins or to clients yet, the value is going somewhere else. Our panel's answer: into quality. And that may be exactly where firms should be making their case.
Featured Content
Who Captures the Value When AI Transforms Patent Practice?
AI is compressing patent workflows in ways that were unimaginable five years ago. Drafting time is down. Prior art searches that once took days take hours. The tools are real, the efficiency gains are measurable, and the profession has largely moved past debating whether to use them.
But a harder question has taken its place: when AI makes patent work faster, where does the value go? Do efficiency gains flow back to clients as lower fees? Do they strengthen firm margins? Do they unlock better quality work that neither side was previously getting? Last week, we convened a panel with Gene Quinn and IPWatchdog to work through exactly that—with five practitioners who are living this question every day.
What emerged wasn’t a clean answer. It was something more useful: a shared framework for thinking about what AI actually does well in patent practice, where human judgment remains irreplaceable, and how firms and clients can stop talking past each other on pricing.
Why This Matters
Quality is winning the debate, but not by accident: When we polled 135 attendees on what matters most when choosing an AI patent platform, 76% said quality of output. Not time savings, not cost reduction. The practitioners delivering quality gains with AI aren’t doing it by stepping back—they’re doing it by staying more deeply involved, using AI to do the work that economics had previously made impossible to do routinely.
Patent preparation and prosecution remains the biggest bottleneck: 45% of attendees cited it as their top workflow challenge. AI is changing the shape of that bottleneck—not eliminating it. Invention disclosures that once arrived as two sentences now arrive as 400-paragraph AI-generated documents, and the real work of identifying the inventive concept hasn’t gone away. It’s just moved.
The practitioner’s expertise is what makes AI output usable: AI doesn’t naturally produce the language needed to meet the Federal Circuit’s §101 test, build litigation-resilient claims, or anticipate examiner behavior. Attorneys who are capturing real value from these tools aren’t delegating to them—they’re directing them, with the end game in view from the first draft.
Adoption at scale is a different problem than individual adoption: A practitioner with a computer science background picking up a new tool is one thing. Deploying that capability across a team of 50 attorneys—with different technical backgrounds, different workflows, and existing document management systems—is another problem entirely. The tools that move organizations are the ones that integrate where practitioners already work.
The inventorship question is becoming practical, not theoretical: As AI becomes more embedded in claim drafting, deposition transcripts in future litigation will ask whether AI generated the claims. Inventors must be able to demonstrate human contribution to conception under penalty of perjury. This isn’t a reason to avoid AI in prosecution—it’s a reason to ensure a practitioner is driving the process at every stage.
The panel’s consensus wasn’t that AI will upend the economics of patent practice overnight. It was that the profession is at an inflection point where the firms and clients who align early on expectations—what AI does, what quality requires, and what the work is actually worth—will be the ones best positioned for what comes next.
That alignment starts with the kind of conversation last week’s session was built around. And it continues in our next session on [DATE], where Romain Vidal and Alex G Lee, PhD, Esq will go deeper on what an integrated AI patent workflow actually looks like in practice—from patentability research through drafting, prosecution, and beyond.
In Case You Missed It
A few complementary reads from past issues and the DeepIP catalog:
▶ Webinar Replay: How to Build an Integrated AI Patent Workflow That Scales
Earlier this week, we invited Alex G Lee, PhD, Esq, Founder & Chief Instructor of the AI-Native Patent Practice Academy and Patent Attorney at TechIPm, to discuss that difference and show what integrated patent AI looks like in practice on the DeepIP platform. Watch the replay now.
▶ Article: Ashwanth Sridhar on the AI Maturity Gap Holding Patent Teams Back
A patent engineer turned manager reflects on why AI tools are advancing faster than the workflows built around them—and what needs to change.
▶ Guide: The Patent Professional’s Guide to Winning Over AI-Skeptical Clients
In-house clients pushing back on AI? This guide gives patent attorneys the talking points and data to address concerns around security, quality, and billing.
See you next week.
The DeepIP Team
Behind the Claims