AI reports: signals that predict PAE activity (are you at risk)

Patent strategy has long been shaped by instinct. Jon Liu, Founder & CEO of ArcPrime, explains how data is changing that and what it means for managing PAE risk. He shares a data-driven perspective on how companies can better understand and manage that risk, from identifying whether your business is likely to be targeted, to understanding which patents create real exposure, and how to think strategically about building a portfolio through filing, acquisition, or both.

Late last year, LOT Network introduced our own dedicated editorial channel, insightsHUB. We made this decision because our community consistently asked for more education, more insights – centralized resources on topics and news related to our community that would quickly offer material to stay current and educated on the market. 

We have published expert analysis on topics as wide-ranging as best practices in patent monetization to an overview of the probable threats that USPTO pro-patent policies hold for crypto, to an informative, deep dive educational multi-part series on PTAB.

But this piece, today’s new, exclusive Q&A with Jon Liu, founder of ArcPrime, is a standout. 

Q: Jon, you bring a unique perspective to this conversation. Today, you’re the founder of an AI-powered platform for patent lifecycle management. But before that, you were on the inside, leading patent analytics at Meta and earlier practicing at Fish & Richardson. What motivated your move from advising on patents to building technology for managing them?

Jon: By the time I started ArcPrime, I had seen firsthand how patent portfolios were often built. Even at the most sophisticated companies, decisions were frequently driven by instinct, by very experienced attorneys making what were essentially educated guesses.

The reason wasn’t a lack of expertise, far from it. The problem was a lack of usable data. Historically, there simply wasn’t a way to analyze enough long-term information to truly understand the value or risk of a patent, much less an entire portfolio. So companies would spend millions, collectively billions across the industry, building portfolios and hoping their assumptions were right.

I saw this directly when managing patent analytics at Meta. One of the hardest questions for any IP leader is understanding risk before it materializes. For example, are you actually a target for a patent assertion entity, or are you just assuming you might be? Historically, you only found out once litigation started, and by then it’s already too late to adjust strategy.

Part of the challenge is the feedback cycle in patents. You usually only see how a patent is used when it’s asserted in litigation or when it’s licensed or sold. Litigation often happens ten years after filing; licensing may take six to ten years. That makes it incredibly difficult for companies to learn from outcomes and improve their decision-making in real time.

What’s changed recently is the ability to analyze those long time horizons at scale. With AI and deeper data analysis, we can now surface patterns in patent activity, where assertion risk is emerging, how portfolios are actually being used, and what signals predict value.

That shift makes it possible to move patent strategy from intuition to evidence. ArcPrime was built around that idea, giving IP leaders the data and analytical tools they need to make more informed decisions about what to file, acquire, prioritize and what to let go.

For me, the opportunity was to bridge two worlds I knew well: the people responsible for managing patent portfolios, and the technology that can now give them much better visibility into the decisions they’re making.

Q: With all the focus on changes at the USPTO and in the current administration, how much should companies actually adjust their patent strategy and how much of that is just noise compared to the underlying data?

Jon: Policy changes and shifts at the USPTO are important, but they tend to operate on a much shorter time horizon than the lifecycle of a patent.

What we focus on in our analysis are long-term patterns, where assertion activity has historically concentrated, how PAEs select assets, and how those behaviors evolve over time. Those trends play out over decades, not election cycles or administrative changes.

From that perspective, the fundamental drivers of patent value and risk are relatively stable. The types of technologies that attract assertion, the characteristics of patents that create leverage, and the strategies used by sophisticated actors, those don’t change quickly in response to policy shifts.

For companies, that means the core of their patent strategy should remain grounded in those longer-term signals rather than reacting to near-term changes. If you’re building a portfolio to protect your business, you want to base those decisions on durable patterns in how patents are actually used, not on variables that may shift again in a few years.

Where policy does matter is at the margin, it can influence timing, procedural strategy, or specific tactical decisions. But it shouldn’t fundamentally change how you think about portfolio construction.

The advantage now is that with better data and analysis, you can separate those long-term signals from short-term noise and make decisions with a much clearer understanding of what actually drives outcomes.

Q: How does AI actually help analyze and predict PAE risk? And why is that kind of insight valuable for in-house counsel managing patent portfolios?

Jon: We started with a simple observation: patent assertion entities are highly rational actors. Their entire business model is built around extracting value from patents, so they’re extremely disciplined about the assets they pursue. They don’t spend time or money asserting weak positions.

That makes their behavior incredibly informative. By studying which patents PAEs choose to acquire, assert, and license, and just as importantly, which ones they ignore, you start to see patterns about what actually creates leverage in the market.

Using AI, we can analyze those patterns across very large datasets and long time horizons. For example, in our recent analysis of over 6,500 patent litigations, we see consistent structural signals, asserted patents tend to come from larger families, are more likely to include continuations, and are more often acquired than built. 

At a more granular level, even claim structure matters. In one dataset comparing nearly 10,000 patents, asserted patents tended to have shorter independent claims, about 150 words on average versus 167 in the control group, but richer underlying disclosures. 

Individually, these signals might seem subtle. But when you analyze them together at scale, they begin to form a predictive picture of which patents are more likely to be asserted or licensed.

For in-house counsel, that’s extremely valuable. When you’re building a portfolio for defensive purposes, understanding what sophisticated monetization actors consistently select for gives you a much clearer framework for what to prioritize, whether that’s how you draft, where you invest, or how you shape families over time.

The other piece is risk management. Many companies feel they might be targets for PAE litigation, but historically, it’s been very difficult to quantify that risk. If you can’t measure it, you’re essentially managing by intuition.

Once you can actually model the probability of being targeted, based on how your portfolio aligns with these real-world patterns, the conversation changes. IP leaders can prioritize more intelligently, allocate resources with confidence, and explain their strategy in a way that’s grounded in data rather than speculation.

Q: You’ve done deep analysis across multiple industries, and we’ll point readers to those individual reports for the specifics, but are there overarching trends that hold true broadly? What are the top three things in-house counsel should know regardless of their sector?

Jon: Yes, when you look across the data, three patterns show up consistently.

The first is the importance of continuations. Patent practitioners have always understood that continuations provide flexibility, they allow you to refine claims as markets and technologies evolve. What’s interesting is that the data strongly reinforces that intuition. In our analysis, roughly 50% of asserted patents include continuations, compared to about 20% in control groups. 

That’s a significant difference, and it reflects how sophisticated monetization actors consistently gravitate toward patents that have been actively shaped over time. For in-house teams, that means a thoughtful continuation strategy isn’t just good practice, it’s empirically tied to how patents ultimately create leverage.

The second is where your technology sits in its lifecycle. An invention moves through a curve, from emerging to mature to well understood. What we see is that assertable cases tend to concentrate earlier in that lifecycle, when the technology is still developing. At that stage, there’s typically less prior art and more room to secure meaningful claim scope.

You can also see this in how patents are structured. Asserted patents tend to have cleaner, more targeted claims, around 150 words on average, versus 167 in non-asserted patents, but with richer disclosures behind them. That combination tends to map more effectively to real products, particularly in emerging areas where standards and implementations are still evolving.

The third insight is about portfolio construction. Many companies assume that building a portfolio organically is the primary path. But the data shows that assertion is often driven by assets that were acquired rather than internally developed. In fact, about 55% of asserted patents were acquired, compared to roughly 38% in control portfolios. 

That suggests acquisition isn’t just a supplement, it’s a core part of how enforcement-relevant portfolios are actually built.  Identifying targeted assets that fill specific gaps can be both efficient and strategically important, especially when others in the market are actively optimizing for those same opportunities.

Across all three, the common thread is that the patents that matter aren’t accidental, they’re shaped.  What’s changing now is that AI allows you to analyze these dynamics much more precisely. Instead of relying on broad assumptions about where value might exist, you can identify which structures, technologies, and strategies are consistently associated with real-world outcomes, and build your portfolio accordingly.

Q: What should readers do if they want to know more about their industries in particular, or about the trend lines you’ve identified? 

ArcPrime has published a 5-part series analysis, available to all LOT Network members via this link.

Jon Liu, Founder & CEO, ArcPrime
Jon Liu is the CEO and founder of an AI-native platform redefining how companies optimize patent portfolios across the full lifecycle – from pre-filing through post-issuance. Prior to founding ArcPrime, Jon spent seven years at Meta as Associate General Counsel, where he managed AI and AR/VR patent portfolios, led enterprise-wide trade secret programs, and developed data-driven approaches to patent strategy. His work included published research on using AI to right-size patent portfolios. Earlier in his career, Jon was a patent attorney at Fish & Richardson, where he drafted foundational patents for Google’s Tensor Processing Unit (TPU) and TensorFlow technologies. Jon is also a frequent speaker and educator on patent and trade secret strategy, having taught courses to hundreds of students on topics ranging from patent portfolio ROI to drafting patents from scratch. He earned his J.D., on full scholarship, from Santa Clara University School of Law and his B.S. in Electrical Engineering and Computer Science from UC Berkeley, where he was a member of Eta Kappa Nu. He is admitted to the California State Bar and registered to practice before the USPTO.