Japan's patent office built an in-house AI system that re-ranks prior-art search results and auto-classifies incoming filings, backed by a nationwide network of AI-trained examiners grown from 13 to 39 since 2023 — though JPO reports no independent efficiency figures yet.
13 to 39 examiners
AI Examination Support Team size (as of 1 October 2023)
40+
Published examination case studies (AI/IoT inventions)
Details
Maturity
Scaling
Promoter
Japan Patent Office (JPO)
Period
2021-2025
Keywords
intellectual property, patent examination, government digitalisation, AI-assisted search
Context
The Japan Patent Office (JPO) has developed in-house AI tools within its 2022-2026 AI Action Plan: Concept Search, which re-ranks prior-art search results by citation likelihood using a machine-learning model, and Patent Classification, which automatically assigns FI/F-term codes to non-Japanese patent documents. On 1 October 2023, JPO grew its AI Examination Support Team from 13 to 39 experts, assigning at least one to every examination office, and on 1 April 2024 created an external 'AI advisor' role. The office has published over 40 examination case studies on AI/IoT-related inventions and manga-illustrated public guidelines explaining patentability criteria.
Implementation
Indicative cost
Medium (€50k–€500k) — No budget figure is published; classified as medium cost, reflecting a dedicated in-house examiner team and multi-year national action plan without disclosed spend.
Time to results
Long (> 3 years) — Runs within a multi-year 2022-2026 AI Action Plan, with the examiner team scaled up in 2023 and an advisor role added in 2024.
Staffing & skills
AI Examination Support Team (grown from 13 to 39 examiners, one per examination office), External 'AI advisor' role (created 1 April 2024)
Conditions for success
Dedicated, growing examiner network embedded in every examination office
Published case studies and public guidelines to build shared understanding of AI-related patentability
Operates within a defined multi-year (2022-2026) AI Action Plan
Common failure modes
Some prior-art search modules remain in agile-development or proof-of-concept phase per JPO's own December 2025 WIPO briefing
No independently audited time-saved or accuracy figures have been published
Commonly funded by
National / regional programmes
Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.
Do you run this practice?
Claim it —
verified implementers get a public contact pathway and can propose corrections.
Data sources
Where this practice's information was retrieved from, and when.