
エージェント型AIとは、知覚、計画、ツールの使用、自己修正のループを反復することで、各意思決定ポイントで人間の介入を必要とせずに、複数のステップからなる目標を自律的に追求するAIシステムを指します。これは、単一ターンの即時応答モデルから、複雑な目標をサブタスクに分解し、外部環境に対してアクションを実行し、結果を評価し、それに応じて計画を修正するアーキテクチャへと構造的に移行するものです。
標準的なエージェントアーキテクチャは、大規模な言語モデル推論コアと、Webブラウザ、コードインタプリタ、APIクライアント、ファイルシステム、データベースインターフェースなどのツールレジストリ、そして短期的な作業コンテキスト、過去のアクションのエピソード記録、取得された長期的な知識を網羅するメモリアーキテクチャを組み合わせ、エージェントが単一のコンテキストウィンドウを超える広範な相互作用の期間にわたって一貫した目標追求を維持できるようにします。マルチエージェント構成では、これをさらに拡張し、オーケストレータによって調整される専門エージェント間でサブタスクを分散させ、エージェント間の連携を導入します。 コミュニケーション プロトコル、結果の集約、および競合解決は、追加のアーキテクチャ上の懸念事項である。
The publications and patents indexed below address planning algorithms, tool use architectures, memory systems, multi-agent coordination protocols, agent evaluation benchmarks, and safety constraint 方法論.
これは、エージェント型AIに関する英語の世界中の出版物と特許の最新のセレクションです。多数の科学オンラインジャーナルの中から、エージェント型AI、AIエージェント、自律型AIエージェント、マルチエージェントシステム、AIエージェントオーケストレーション、エージェントプランニングループ、ReActエージェント、思考連鎖プランニング、ツール使用型AIエージェント、関数呼び出しAI、AIエージェントメモリ、長期エージェントメモリ、エピソードエージェントメモリ、エージェントワーキングメモリ、AIエージェントタスク分解、階層型エージェントプランニング、AIエージェント自己修正、AIエージェント反射、AIエージェント評価、エージェント報酬形成、AIエージェントサンドボックス、エージェント安全ガードレール、マルチエージェント協調、エージェント通信プロトコル、AIエージェントツールレジストリ、コード実行エージェント、ウェブブラウジングエージェント、検索拡張エージェント、エージェントベンチマーク評価、ヒューマンインザループエージェントに分類され、焦点を絞っています。
Ai agent for core network
Patent published on the 2026-10-15 in WO under Ref WO2026213740 by NOKIA TECH OY [FI] (Gajic Borislava [de], Sun Shu Qiang [cn], Abdelkader Abdelrahman [de], Khatibi Sina [de], Schirilla Achill Andreas [de])
Abstract: Example embodiments of the present disclosure are directed to artificial intelligence (AI) agent for core network. A first network entity receives, from a second network entity or an apparatus, a request originated from the apparatus for an agentic service. The first, second network entities and the apparatus implement an AI agent. The first network entity determines a plan for the agentic service at least based on capability information of at least one device. The plan is associated with one or[...]
Our summary: An AI agent for core networks processes requests for agentic services. It determines a service plan based on device capabilities and executes associated tasks. The execution status is communicated back to the requesting entity.
AI agent, core network, agentic service, execution status
Patent
Ai agent and generative ai-based b2b solution comparison, recommendation, and deployment automation system and method
Patent published on the 2026-10-08 in WO under Ref WO2026212264 by IMPACT FLOW INC [KR] (Ryu Hyokwon [kr])
Abstract: Disclosed is a B2B solution deployment automation apparatus and a method that use an AI agent and a generative AI model to generate user requirement-based B2B solution recommendations and comparison reports, and furthermore, automatically orchestrate a deployment process such as integration configuration, API connection, and data migration, and provide operational optimization feedback through calculation of performance, cost, and security indices.[...]
Our summary: The system automates B2B solution deployment using an AI agent and generative AI model. It generates user requirement-based recommendations and comparison reports. The method also orchestrates integration, API connections, and data migration while providing optimization feedback.
B2B solutions, AI agent, generative AI, deployment automation
Patent
Secured distributed network context management for ai agent
Patent published on the 2026-10-08 in WO under Ref WO2026209409 by NOKIA TECH OY [FI] (Aggarwal Chaitanya [de], Gajic Borislava [de], Tang Haitao [fi], Mishra Rajendrakumar Premnarayan [in])
Abstract: Example embodiments of the present disclosure are directed to secured distributed network context management for an artificial intelligence (AI) agent. A method comprises verifying, based on a context distribution policy associated with a second apparatus comprised in the at least one apparatus, that a network context is authorized to be provided to the second apparatus, wherein the context distribution policy indicates identifier information of the second apparatus and an indication of a networ[...]
Our summary: The method verifies network context authorization based on a context distribution policy. It identifies the second apparatus and the allowed network context. The authorized context is transmitted to the second apparatus for a generative AI agent.
secured distributed network, context management, artificial intelligence, policy verification
Patent
Collaborative and cooperative processing method for tasks in wireless network system, wireless network system, electronic device and computer program
Patent published on the 2026-10-08 in WO under Ref WO2026208347 by ZTE CORP [CN] (Yang Li [cn], Sun Wenwen [cn], Xie Feng [cn])
Abstract: Provided in the embodiments of the present disclosure are a collaborative and cooperative processing method for tasks in a wireless network system, a wireless network system, an electronic device and a computer program product, wherein the method is applied to a network element node, a communication-computing service layer and an AI agent (AA) task layer being deployed on the network element node, and the AA task layer comprising one or more AA functional entities. The method comprises: the AA t[...]
Our summary: The method enables collaborative processing for tasks in a wireless network system. It involves interaction between an AI agent task layer and a communication-computing service layer. This approach addresses efficient cooperation between different network element nodes across various vendors.
collaborative processing, wireless network, AI agents, task cooperation
Patent
System and method for accelerating operations of an ai agent on a digital interface
Patent published on the 2026-10-08 in US under Ref US20260311007 by EVINCED INC [US] (Synett Yossi [il], Moav Gal [il], Thadani Navin R [us], Or Evyatar Yaakov [il], Hakman Tair [il], Orgil Kfir Yehuda [il], Bracha Ben [il])
Abstract: [0000] The method and system for accelerating operations of an AI agent is presented. The method includes receiving, by a first AI agent, a task to be performed on at least one digital interface based on a natural language request from a client device; extracting, by the first AI agent, at least one learned skill to be used by the first AI agent for completing the task based on accessibility semantics of at least one component of the at least one digital interface, wherein learned skills are cap[...]
Our summary: The system accelerates AI agent operations on digital interfaces. It involves receiving tasks from client devices and extracting learned skills based on accessibility semantics. The AI agent completes tasks using these skills effectively.
AI agent, digital interface, natural language processing, learned skills
Patent
Novel edge-based multi-agent system for real-time data validation and quality grading
Patent published on the 2026-10-08 in US under Ref US20260312968 by PARA RAGHU KUL THILAK [CA] (Para Raghu Kul Thilak [ca])
Abstract: [0000] present invention relates to an edge-based multi-agent data validation system that executes specialized checks (schema, domain, semantic, anomaly) on incoming data streams in near real time. A coordinator module aggregates validation outputs and assigns a quality tier label (bronze, silver, or gold) to each data unit. This immediate classification prevents contaminated data from infiltrating downstream pipelines, thereby reducing bandwidth usage and storage overhead. Agents autonomously a[...]
Our summary: The invention presents an edge-based multi-agent system for real-time data validation. It performs specialized checks on incoming data streams and assigns quality tier labels. The system enhances data integrity and reduces resource usage in distributed environments.
multi-agent system, data validation, real-time processing, edge computing
Patent
A Scoping Review
Published on 2026-02-01 by Jonathan Gibson, Praveen Chinniah, Shashank Chapala, Ojasvi Vemuri, Rajesh Botchu @MDPI
Abstract: Objectives: Artificial intelligence (AI) is a transformative development in the field of medicine. In the field of musculoskeletal radiology, agentic AI is a technology that could flourish, but currently, the limited evidence base is fragmented and sparse, and we present a scoping review of it. Methods: Parallel searches were conducted in four databases: PubMed, Embase, Scopus, and Web of Science. Search terms included all agentic AI and autonomous AI agents, as well as radiology. All papers und[...]
Our summary: This scoping review evaluates the potential of agentic AI in musculoskeletal radiology. It identifies eleven relevant studies highlighting improved decision support, workflow optimization, and image analysis. Despite promising findings, the evidence base remains limited and theoretical.
AI in Radiology, Musculoskeletal Imaging, Workflow Optimization, Decision Support
Publication
A Novel Security Index for Assessing Information Systems in Industrial Organizations Using Web Technologies and Fuzzy Logic
Published on 2026-01-29 by Sulieman Khaddour, Fares Abu-Abed, Valery Bogatikov @MDPI
Abstract: Industrial information systems based on web technologies (ISOWT) face escalating security challenges, particularly in critical sectors such as energy. Traditional qualitative security assessments often lack the ability to deliver actionable, real-time insights for managing complex, dynamic threats. This paper proposes a novel security index for evaluating ISOWT in industrial organizations by integrating fuzzy logic, metric-based evaluation, fuzzy Markov chains, and multi-agent systems. The propo[...]
Our summary: This paper introduces a novel security index for evaluating industrial information systems using web technologies and fuzzy logic. It quantifies deviations from an ideal safety center, enabling early risk detection and proactive mitigation. The methodology shows significant improvements in security metrics and outperforms existing assessment techniques.
Security Index, Fuzzy Logic, Information Systems, Web Technologies
Publication

















