|
2026年9月2日付の日経クロステックの記事「オムロンがBedrock活用で知財AIエージェント内製、特許関連工数を50%減」によれば、オムロンは、AWSの生成AI基盤「Amazon Bedrock」などを活用し、先行技術調査と発明説明書作成を支援する知財AIエージェントを自社開発し、2026年1月頃から研究開発部門で利用を始め、これらの業務に要する工数を、少なく見積もっても50%削減したということです。 オムロンでは、発明考案のプロセスを、①発明の創出、②先行技術調査、③発明説明書の作成、の3段階に整理しています。エンジニアが本来注力したいのは発明の創出ですが、専門的な特許文献の確認や説明書作成に多くの時間を取られていました。 そこで、特許庁から取得した過去の出願データをAmazon S3に格納し、Amazon OpenSearch ServiceとAmazon Bedrockを組み合わせました。AIエージェントは、入力されたキーワードや仕様書を基に関連文献を検索・要約し、類似技術の整理、競合分析、発明説明書の作成まで支援します。 商用データベースやSaaSの利用も検討しましたが、データ量の制約や独自の業務プロセスへの対応を考え、内製を選択しました。また、出願前の機密情報を安全に扱うため、AWS上に共通AI基盤「RDinX」を構築し、セキュリティー設定やAIエージェントの挙動、プロンプトを一元的に管理しています。 今回の事例は、生成AIの活用が単なる文書要約から、発明発掘、調査、分析、説明書作成をつなぐ業務プロセスの再設計へ進んでいることを示しています。AIに判断を任せ切るのではなく 、人が結果を確認しながら、エンジニアを定型作業から解放し、発明創出に集中させる取り組みとして注目されます。 この記事の内容を生成AIに深掘り調査させましたのでご参照ください。なお、生成AIによる調査・分析結果は、公開された情報だけに基づくものであり、必ずしも実情を示したものではなく、誤った情報を含む可能性があることにご留意のうえ、ご参照ください。 オムロンがBedrock活用で知財AIエージェント内製、特許関連工数を50%減 2026.09.02 https://active.nikkeibp.co.jp/atcl/act/19/00798/080300002/ OMRON Develops In-House IP AI Agent, Cutting Patent-Related Workload by 50% According to a Nikkei XTECH article published on September 2, 2026, titled “OMRON Develops In-House IP AI Agent Using Bedrock, Cutting Patent-Related Workload by 50%,” OMRON has developed its own intellectual property AI agent using Amazon Web Services’ generative AI platform, Amazon Bedrock, and other AWS services. The agent supports prior-art searches and the preparation of invention disclosure documents. OMRON began using it within its research and development divisions around January 2026 and has conservatively estimated that it has reduced the time required for these tasks by at least 50%. OMRON divides the invention development process into three stages: (1) generating an invention, (2) conducting a prior-art search, and (3) preparing an invention disclosure document. Although engineers would ideally devote most of their attention to creating inventions, they had been spending considerable time reviewing highly technical patent documents and preparing disclosure documents. To address this issue, OMRON stored historical patent application data obtained from the Japan Patent Office in Amazon S3 and combined Amazon OpenSearch Service with Amazon Bedrock. Based on keywords or technical specifications entered by users, the AI agent searches for and summarizes relevant documents. It also assists with organizing similar technologies, analyzing competitors, and preparing invention disclosure documents. OMRON considered using commercial databases and software-as-a-service solutions. However, it chose to develop the system in-house because of restrictions on data volume and the need to accommodate its proprietary business processes. To securely handle confidential information before patent applications are filed, the company also built a shared AI platform called “RDinX” on AWS. The platform centrally manages security settings, AI agent behavior, and prompts. This case illustrates how the use of generative AI is evolving beyond simple document summarization toward the redesign of entire workflows connecting invention discovery, prior-art research, analysis, and the preparation of disclosure documents. Rather than leaving decisions entirely to AI, OMRON has adopted a human-in-the-loop approach in which people review the AI-generated results. The initiative is particularly noteworthy because it frees engineers from routine tasks and enables them to concentrate more fully on creating inventions. I asked generative AI to conduct an in-depth investigation of the matters discussed in this article. Please refer to the results below. Please note, however, that the research and analysis generated by AI are based solely on publicly available information. They may not necessarily reflect the actual circumstances and may contain inaccurate information. Your browser does not support viewing this document. Click here to download the document. Your browser does not support viewing this document. Click here to download the document. Your browser does not support viewing this document. Click here to download the document. Your browser does not support viewing this document. Click here to download the document. Your browser does not support viewing this document. Click here to download the document. Your browser does not support viewing this document. Click here to download the document. Your browser does not support viewing this document. Click here to download the document. Your browser does not support viewing this document. Click here to download the document. Your browser does not support viewing this document. Click here to download the document.
0 Comments
Leave a Reply. |
著者萬秀憲 アーカイブ
April 2026
カテゴリー |
RSS Feed