Preliminary Sneak Preview of Confirmed Speakers

Patent Analytics Integrated with Authoritative Scientific Data & AI

Stephen K Boyer - Molecular Assets, Collabra Inc.

Transformer-based large language models (LLMs) such as ChatGPT and Gemini generate responses by predicting statistically likely text continuations. While effective for general language tasks, these models can be unreliable for patent intelligence workflows, where rare technical relationships and precise attribution matter. In such settings, transformer LLMs may produce confident-sounding but incorrect statements (“hallucinations”), particularly when asked to infer uncommon associations across compounds, targets, mechanisms, or claims.

We will present a hybrid LLM designed as an authoritative alternative for patent search and analysis. Our system is grounded in validated scientific facts derived from hundreds of structured database tables and integrates billions of curated scientific entities extracted from worldwide patents and the broader scientific literature. This architecture supports high-precision querying across patents, inventors, organizations, compounds, targets, and biological relationships—while maintaining explicit evidence for frequent relationships and robust performance on rare relationships. The result is a patent-focused AI system that delivers traceable, fact-grounded outputs without hallucination.

Fantastic Platform Patents and where to find them

Daniela Hana Bar El & Yonit Ronen – D&Y IP

Platform patents comprise claims directed to a biotechnological system or process that may relate to multiple different products. When conducting Freedom-to-Operate (FTO) searches, it is essential to identify such platform patents, as they typically provide broad protection covering a class of products and may impact entire product pipelines rather than a single product candidate.

This presentation addresses the critical need for specialized search strategies to effectively identify and evaluate these dominant patents. Standard keyword and classification-based searches are often insufficient, as they may fail to capture the full scope of these patents.

•   We explore advanced, multi-faceted methodologies, including bio-sequence, chemical, and artificial intelligence search strategy designs.

•   Using a multi specific Chimeric Antigen Receptor (CAR) technology as a representative case study, we demonstrate how to design a comprehensive search strategy that integrates multiple technological elements, including textual analysis, bio-sequence searching, and chemical structure searching.

•   Finally, we generalize the insights gained from this example to propose a structured framework for searching complex biotechnological platforms. Key guiding questions for strategy design are identified, providing practical guidance for patent professionals conducting robust and reliable FTO searches.

Human IP Experts + Value-added databases = Human Intelligence

Matt Eberle & John Willmore - BizInt Smart Charts

PIUG was founded in part to encourage patent data publishers to make their databases more useful for patent searchers. In the age of AI, getting some patent information is arguably easier than ever. What then, is the value of value-added databases and how do we leverage it?

Does it matter which IP you find and what details you can determine? Human expertise with value-added sources makes for better intelligence.

Patent sources are not identical whether in coverage or content. We’ll look at some example searches carried out on both free and value-added databases and analyze the results to see what documents we find in multiple sources and which in a single source. And, we'll look at how finding the same IP in multiple sources makes for a richer view of the landscape thanks to unique fields (and human editors).

What about AI generated content? Human expertise and the transparency provided by value-added databases offer the best way forward.

A Strategic Two-Pronged Approach to Freedom to Operate (FTO) for Complex Formulations

Harita Achanta – SciTech Patent Art Services Pvt. Ltd.

Conducting Freedom to Operate (FTO) analyses is a critical milestone of a product development cycle. Given the sensitive nature of such analyses where the full product formulation needs to be reviewed and searched, companies are often sensitive to outsourcing such search needs. However, when there are frequent new product launches or when internal resources are limited, the need to outsource such searches is high. IP or R&D professionals, in such situations, face a strategic dilemma of how to conduct exhaustive patent clearances externally without compromising sensitive technical know-how or overextending budgets.

This presentation outlines a well demonstrated, two-pronged methodology for conducting FTO searches on client-defined formulations, while effectively balancing confidentiality and cost, in order to overcome enforcement barriers.

We contrast two distinct strategies for FTO:

1.    The Complete Formulation Search:

This comprehensive approach, used when confidentiality constraints are minimal, yields a highly refined set of probable blocking patents suitable for direct decision-making.

2.    The Isolated Ingredient Search:

When confidentiality is critical, the formulation is deconstructed and searched by individual ingredients, while the final synthesis of results is performed by the client.

Attendees will gain insights into the specific query structures, and outcomes associated with each approach. By the end of the session, participants will be equipped to select the most appropriate approach based on their budgetary constraints, and the sensitivity of their chemical or biological innovations.

Baseline IP Knowledge for Clients

Matthew Holley - Pharmaceutical Officers Academy

Presentation of Base Term, Patent Term Adjustment, Patent Term Extension (touching on international equivalents), and Pediatric Extension with a PowerPoint deck that will be made available to the audience to use for them to make the presentation to their clients.  Or, their clients can take the course for free on the Pharmaceutical Officer Academy website.  Discussion to focus on how to leverage such shared baseline knowledge into good communications with clients and within corporate research and legal functions.

Freedom to Operate Studies assisted by Large Language Models

Zhifu Shu - ExxonMobil Technology and Engineering Company

Freedom to operate (FTO) addresses a critical question for commercialization: can a product be made, used, sold, or deployed in a target market without infringing on others’ rights?  As patent landscapes become increasingly complex and global, FTO must be completed faster without sacrificing quality or rigor. This presentation explores how large language models (LLMs) can support FTO in two key ways.  First, as a review tool, LLMs can help speed up the screening of search results by following human guidance on what to keep or discard. 

Second, as a search tool, LLMs can enhance coverage by suggesting synonyms, broader concepts, and relevant classifications that complement traditional search methods.  We will share examples of search performance using both commercial and in-house models.

Finally, we will discuss insights from early testing—where LLMs consistently add value, where they fall short, and the governance practices that ensure safe adoption, including human checkpoints and uncertainty tagging.