Science, Technologies, Innovations №3(39) 2026, 123-128 р

https://doi.org/10.35668/2520-6524-2026-3-14

Havrylei L. M. — Ukrainian Patent Attorney, patent search specialist, Head of the PatentProm Kyiv Patent and Legal Project; 3, Yevhena Hutsala Lane, P.O. Box 55, Kyiv, Ukraine, 01011; +38 (068) 366-07-72; gavrileyl@gmail.com; ORCID: 0009-0001-0903-5432

PATENT SEARCH WITH ARTIFICIAL INTELLIGENCE: X/Y/A TOOLS AND THE “6P” SYSTEM

Abstract. The article examines the transformation of professional patent information search as artificial intelligence tools become integrated into official patent infrastructure. The topic is particularly relevant because in 2026 WIPO introduced PATENTSCOPE AI-Assisted Search, which converts natural-language instructions into structured patent search queries, while the European Patent Office expanded AI capabilities in examiner tools, including a large-language-model component in PreSearch. The purpose of the study is to develop a public control model that enables practitioners to benefit from AI-assisted search without replacing professional verification with automatically generated conclusions. The methodological basis combines the PCT citation categories X, Y and A, WIPO requirements for documenting the search process, current EPO materials on the use of AI, and recent research on semantic patent retrieval. The paper proposes a functional integration of X/Y/A categories, which describe the patent relevance of a document, with the author’s “6P” system (Ukrainian designation “6П”), defined as a six-stage control system covering provenance, the primary source, verification of evidence against the claim, disclosure of machine transformations, reproducibility of the search protocol, and personal and institutional responsibility. The analysis shows that these two layers are not interchangeable: correct categorisation does not eliminate the need to verify the exact disclosure of a critical feature, while careful verification of a document does not by itself determine its significance for novelty or inventive step. Particular attention is paid to negative search results. A conclusion that a document was not found must remain limited to the databases, languages, classifications, queries, jurisdictions, dates and technical conditions actually covered by the search; therefore, “not found” cannot be equated with “does not exist”. The proposed model separates machine retrieval, evidentiary verification and professional judgment. It may be applied to prior-art searching, novelty and patentability assessment, patent analytics and quality control of AI-assisted patent search results while preserving human oversight and responsibility.

Keywords: patent search, patent information, artificial intelligence, prior art, X/Y/A, “6P” system (“6П”), novelty, inventive step, verification, primary source.

REFERENCES

  1. (2026, July 2). PATENTSCOPE AI-Assisted Search Now Available. World Intellectual Property Organization. Retrieved from: https://www.wipo.int/en/web/patentscope/w/news/2026/patentscope-ai-assisted-search-now-available.
  2. (2026). PCT International Search and Preliminary Examination Guidelines (Chapter 16, paras. 16.65–16.69). World Intellectual Property Organization. Retrieved from: https://www.wipo.int/en/web/pct-system/texts/ispe/16_22_85.
  3. (2026, January 1). PCT International Search and Preliminary Examination Guidelines (Chapter 21, para. 21.26). World Intellectual Property Organization. Retrieved from: https://www.wipo.int/en/web/pct-system/texts/ispe/21_26.
  4. (2022). WIPO Guide to Using Patent Information. World Intellectual Property Organization. Retrieved from: https://www.wipo.int/publications/en/details.jsp?id=4615&plang=EN.
  5. (2026). Guidelines for Examination in the European Patent Office: General Part, section5, The use of artificial intelligence. European Patent Office. Retrieved from: https://www.epo.org/en/legal/guidelines-epc/2026/foreword_5.html.
  6. (2026, June 30). EPO publishes Annual Review 2025. European Patent Office. Retrieved from: https://www.epo.org/en/news-events/press-centre/press-release/2026/1362807.
  7. Artificial intelligence: AI in EPO tools. European Patent Office. Retrieved from: https://www.epo.org/en/news-events/in-focus/ict/artificial-intelligence.
  8. Androshchuk, H. O., & Khvostenko, V. S. (2026). Patentnyi landshaft tekhnolohii Visual SLAM: stratehichni naslidky dlia avtonomnykh system [Patent landscape of Visual SLAM technologies: Strategic implications for autonomous systems]. Nauka, tekhnolohii, innovatsii [Science, technology, innovations]. 1(37), 42-57. Retrieved from: https://doi.org/10.35668/2520-6524-2026-1-05 [in Ukr.].
  9. Chikkamath, R., Rastogi, D., Maan, M., & Endres, M. (2024). Is your search query well-formed? A natural query understanding for patent prior art search. World Patent Information, 76, 102254. DOI: https://doi.org/10.1016/j.wpi.2023.102254.
  10. Chikkamath, R., Andersson, L., & Endres, M. (2026). Rethinking patent retrieval with language models: Toward scalable and efficient search. World Patent Information, 84, 102433. DOI: https://doi.org/10.1016/j.wpi.2026.102433.
  11. Yoshikawa, N., & Krestel, R. (2025). Do large language models understand patents? Enhancing patent classification through AI-generated summaries. World Patent Information, 81, 102353. DOI: https://doi.org/10.1016/j.wpi.2025.102353.
  12. Bui, L. V. (2025). Advancing patent law with generative AI: Human-in-the-loop systems for AI-assisted drafting, prior art search, and multimodal IP protection. World Patent Information, 80, 102341. DOI: https://doi.org/10.1016/j.wpi.2025.102341.

Received 08/10/2026
Accepted for publication 08/28/2026
Published 09/30/2026

License
Copyright (c) 2026 Science, Technologies, Innovations Journal

All materials published in the current issue of the journal are distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0): https://creativecommons.org/licenses/by-nc/4.0/deed.uk
The license permits free non-commercial use, distribution, reproduction, and adaptation of the materials in any format, provided that proper attribution is given to the author(s) and the source of publication is cited. Commercial use of the materials is permitted only with the publisher’s written consent.