AI, Patent Strategy, and What Actually Drives Outcomes in 2026 – Part 2

By Daniel J. Holmander and Michel “Mike” Morency, Ph.D. on May 12, 2026

Part 2: AI and Patent Prosecution — Why Strategy Still Determines Portfolio Value

Introduction

Part 1 of this series focused on diligence and opinions. Patent prosecution is undergoing the same transition: AI has dramatically improved efficiency, but strategic value increasingly comes from understanding where automation helps, where it creates risk, and where experienced judgment still determines portfolio value.

By 2026, every sophisticated practice is using AI. Specifications can be drafted faster, prior art analyzed more comprehensively, and AI-assisted systems are increasingly useful for identifying support issues under 35 U.S.C. §112, organizing disclosure, and stress-testing claim language before filing. Those are meaningful advances, particularly in biotechnology where written description and enablement remain central concerns after Ariad Pharm., Inc. v. Eli Lilly & Co., 598 F.3d 1336 (Fed. Cir. 2010), and Amgen Inc. v. Sanofi, 598 U.S. 594 (2023). But as baseline drafting becomes more standardized, the quality of prosecution strategy—and its alignment with the underlying science and commercial trajectory—increasingly determines portfolio value.

Patent Applications Now Serve Multiple Stakeholders

Modern life sciences patents are evaluated by investors, strategic partners, licensing groups, diligence counsel, litigators, and potential acquirers—each through a different commercial lens. Business teams focus on exclusivity around the lead product. Litigation counsel evaluates prosecution history and claim construction risk. But investors and partners increasingly ask a more specific question: does the portfolio architecture reflect an understanding of how the technology is likely to evolve, or does it merely protect a snapshot of today’s product?

A portfolio built around a single monoclonal antibody may look adequate at the Series B—but if the continuation strategy does not account for the near-certainty that the platform will extend into antibody-drug conjugates, bispecific formats, or Fc-engineered variants, the portfolio is already aging at the time of filing. Strategic partners evaluating an in-licensing opportunity ask a related question: does the IP cover their likely development pathway? If the portfolio only protects the originator’s exact embodiment, the partner’s exclusivity depends entirely on regulatory data exclusivity and speed to market—a weaker negotiating position.

How Biologics and Platform Technologies Actually Evolve

Prosecution strategy in life sciences is demanding because the underlying science does not stand still. Biologics evolve continuously—often in directions foreseeable scientifically but not prioritized commercially at the time of the original filing.

Monoclonal antibody platforms illustrate this clearly. A company may begin with a conventional IgG1 targeting a validated oncology antigen. But antibody engineering has been moving in well-established directions: Fc engineering to modulate effector function and half-life; conjugation chemistry for ADCs with increasingly sophisticated linker-payload designs; bispecific and multispecific formats for T-cell redirection or multi-epitope engagement; and conditional activation technologies—masked antibodies and protease-cleavable constructs—to improve therapeutic windows. Each represents a distinct competitive pathway with distinct enablement challenges. A specification describing only the parental antibody will not support claims to ADC conjugates or bispecific constructs—even using the same variable regions. After Amgen, the written description and enablement requirements for functionally defined antibody claims are stringent. Prosecution strategy must anticipate platform evolution at the time of original filing and build disclosure depth around foreseeable next-generation formats, or risk losing the ability to claim them later.

Similar dynamics play out across modalities. Cell therapy has evolved from autologous CAR-T toward allogeneic approaches, iPSC-derived products, and in vivo gene-editing strategies—each creating new IP inflection points around construct design, cell processing, and potency assays. mRNA/LNP platforms have branched from vaccines into protein replacement, in vivo gene editing delivery, neoantigen cancer vaccines, and tolerizing constructs—with the LNP formulation itself becoming a competitive battleground. AAV gene therapy involves serotype selection, capsid engineering, promoter design, and transgene optimization—each layer generating distinct IP. These evolutionary trajectories are visible in published literature, competitor filings, and clinical development programs. Experienced prosecutors build disclosure and claim strategies around where the platform is heading, not just where it is today.

The Three Layers of Modern Patent Drafting

The strongest specifications do three things simultaneously: protect the commercial embodiment, cover foreseeable competitive alternatives, and manage disclosure strategically.

At the first layer, AI is genuinely useful—improving drafting consistency, identifying §112 support gaps, and reducing omissions. But products evolve. A biologic entering Phase 1 as a lyophilized IV formulation may emerge from Phase 3 as a subcutaneous liquid in a pre-filled syringe with a different buffer, concentration, and dosing regimen—driven by PK data and market access considerations that did not exist at filing. If the specification does not support those variations, the company may be unable to patent its own commercial product configuration.

At the second layer, effective prosecution anticipates competitive design-arounds. In biologics, competitors engineer around—using different CDR sequences binding the same epitope, different formats, different conjugation strategies, or different modalities directed at the same target. Future competitive pressure often becomes visible in published crystal structures, competitor trial designs on ClinicalTrials.gov, conference presentations, and early-stage patent filings. A prosecutor who tracks these signals can draft specifications that constrain future competitive entry.

At the third layer, poorly structured specifications can unintentionally become R&D roadmaps for competitors. AI tools can now mine published applications at scale, identifying underclaimed disclosure areas that competitors may exploit. For platform technologies with long development timelines—where multiple competitors may be pursuing the same target over seven to twelve years—the decision about what to disclose, when, and in what form can have consequences that persist long after the original filing. Disclosure strategy requires understanding both the patent landscape and the competitive science.

R&D Branching and Prosecution Timing

One of the most underappreciated challenges in life sciences prosecution is that R&D programs branch—sometimes dramatically—during development, and the portfolio must evolve accordingly. An antibody initially developed for non-small cell lung cancer may generate Phase 2 data suggesting unexpected efficacy in an autoimmune indication. The company faces a branching decision: pursue oncology, pivot to autoimmune, or run parallel programs—each with different competitive landscapes, regulatory strategies, and patent needs. The method-of-treatment claims adequate for oncology may not cover the dosing, patient population, or combination regimen relevant to autoimmune. New continuations may be needed, and the window depends on whether the original specification provides adequate written description support.

Manufacturing scale-up creates another branching point often underestimated from an IP perspective. The transition from bench-scale to GMP manufacturing frequently involves changes in cell culture conditions, purification processes, formulation chemistry, and analytical methods—all potentially patentable innovations. Companies that treat manufacturing IP as an afterthought often discover during biosimilar challenges or licensing negotiations that their process patents are their most commercially important assets, because process claims are harder to design around and easier to enforce through supply-chain monitoring. Companion diagnostics and biomarker strategies create parallel prosecution tracks that must be coordinated with the therapeutic portfolio.

The most effective prosecution strategies are explicitly synchronized with clinical and regulatory milestones. The initial provisional is typically filed around IND-enabling data. But as development progresses, each phase generates new filing opportunities. Phase 1 PK data may support specific formulation or dosing claims. Phase 2 dose-response data may enable method-of-treatment claims unsupportable at original filing. Phase 3 subpopulation analyses may identify biomarker-defined populations warranting separate claims. Missing these windows can permanently narrow the portfolio. Lifecycle management—new formulations, devices, indications, delivery devices—extends effective exclusivity, but the foundation for those later filings must be laid years earlier in the original specification’s disclosure.

Sophisticated investors understand this. During diligence, they evaluate continuation chains: How many undocketed continuations remain? Does the specification support claims not yet pursued? Is there prosecution history estoppel narrowing future scope? Are filing dates aligned with clinical milestones, or do they suggest reactive gap-filling? The answers reveal whether the portfolio represents optionality—the ability to pursue future positions as science and market evolve—or merely a static collection of issued patents.

Why Inventor Interviews Matter More in the AI Era

As AI standardizes drafting, the quality of inventor intake increasingly determines portfolio strength. Many of the most valuable aspects of an invention never appear in the disclosure deck—they emerge during substantive conversations. An inventor may mention that a lead antibody’s binding epitope is conformationally sensitive, with profound implications for competitive design-around analysis. A process scientist may describe a chromatographic purification innovation that would never appear in a therapeutics-focused disclosure. A formulation chemist may note unexpected stability results supporting valuable lifecycle claims.

These are not hypothetical examples. They reflect the day-to-day reality of prosecution in the life sciences, where the most strategically valuable disclosure often sits at the intersection of scientific disciplines—and only emerges when the prosecutor has the technical fluency to ask the right questions and recognize the significance of the answers. AI is effective at organizing and stress-testing disclosure once captured. Identifying where future portfolio value is likely to emerge, however, still depends on scientific dialogue, technical instinct, and strategic prosecution experience.

Examiner Interviews Still Shape Outcomes

AI is increasingly effective at modeling examiner behavior and organizing prosecution strategy, but examiner interaction remains deeply human. Practitioners may have less than an hour to advance a complicated discussion, placing a premium on preparation, technical fluency, and credibility. The strongest interviews are rarely adversarial—experienced prosecutors understand when to push, when to narrow, how to frame scientific unpredictability persuasively, and how to preserve meaningful claim scope while advancing prosecution productively. See MPEP §713, which encourages interviews as a mechanism for clarifying issues and advancing efficient examination.

In biologics, demonstrating that a genus of antibodies is not enabled without undue experimentation may require walking through binding data, explaining unpredictable sequence-function relationships, and distinguishing claimed structural features from prior art. That conversation requires both scientific depth and advocacy skill—it cannot be scripted by AI, because the examiner’s real-time reactions shape how the discussion unfolds.

The Growing Problem of AI-Generated Prosecution Responses

The increasing use of AI inside prosecution is creating its own problems. Many examiners now see responses that are lengthy but strategically weak—boilerplate arguments, generic citations, and conclusory assertions that fail to engage the actual rejection. The problem is acute in life sciences, where prosecution arguments depend on specific experimental data, structural biology, or mechanistic reasoning that AI-generated text tends to generalize. An obviousness argument for a biologic may turn on whether a skilled artisan would have had a reasonable expectation of success in combining particular structural modifications—a question requiring specific data, not recited legal standards. The issue is not whether firms use AI. They all do. The issue is whether experienced practitioners are directing the strategy and whether the output reflects genuine scientific engagement.

Conclusion

In 2026, sophisticated life sciences prosecution is no longer simply about drafting applications efficiently. It is about building portfolio architecture that supports future transactions, competitive positioning, enforcement flexibility, and long-term enterprise value—while remaining grounded in the underlying science. AI is now indispensable to that process. It accelerates drafting, improves prior art analysis, and enables more systematic portfolio management. But the strategic decisions that determine portfolio value—how to track platform evolution, when to file continuations, what to disclose and what to reserve, how to anticipate competitive design-arounds, and how to align the patent estate with the branching trajectory of R&D—remain fundamentally human judgments requiring scientific fluency, commercial awareness, and deep strategic experience.

For investors and partners, the quality of those judgments is increasingly visible in the portfolio itself. A well-architected patent estate signals scientific sophistication and commercial foresight. A portfolio that merely protects today’s product in today’s configuration signals the opposite.

Next in this series: how AI is reshaping licensing strategy, transactions, and portfolio valuation—and why sophisticated investors focus on optionality and prosecution architecture rather than raw patent counts.

Disclaimer: This article is for informational purposes only and does not constitute legal advice. The views expressed are those of the authors and do not necessarily reflect the views of their respective firms or clients.