Enterprise Search QuickSolution® Equipped with Search Agent

Sumitomo Electric Information Systems Co., Ltd.'s enterprise search "QuickSolution®" has enhanced its generative AI collaboration (RAG*1) function*2 and is now equipped with an autonomous AI agent, the "Search Agent," that searches and answers. Users can simply ask questions via chat without being conscious of where information is stored. The Search Agent formulates search conditions, searches internal information, evaluates results, reiterates searches as needed to generate precise answers. This significantly reduces the time and effort required for information retrieval with QuickSolution® and extracts the value of internal knowledge.

QuickSolution®, which has achieved the No. 1 share in four market surveys*3, is a purely domestic enterprise search system that has celebrated its 25th anniversary since its release. It supports a full range of up to hundreds of terabytes (TB), allowing cross-sectional searches of information scattered across file servers, various internal systems, and cloud services, down to the content of documents. Furthermore, even scanned image PDFs can be easily found with image OCR search.

Internal information is often dispersed across file servers, various internal systems, and cloud services, leading to trial and error such as "formulating search conditions → searching internal information → evaluating search results → revising search conditions" until the necessary information is obtained. The Search Agent now equipped in QuickSolution® autonomously executes this trial-and-error process as an agent flow, reducing the burden on users.

Search Agent executes search, evaluation, and answer generation in a series

QuickSolution®'s Search Agent repeatedly follows the processes below based on user questions to arrive at precise answers.

1. Formulation of search conditions: Sets search conditions such as keywords from the question

2. Search: Cross-searches internal information across file servers, various internal systems, clouds, etc.

3. Evaluation of search results: Evaluates search results and excludes inappropriate results

4. Answering: If necessary information is gathered, it organizes the key points and generates a precise answer

※ If necessary information is not gathered, the search conditions are revised based on the evaluation results, and the exploration continues.

No prior narrowing down of information sources, Search Agent substitutes for trial and error

Users do not need to narrow down information sources beforehand and do not need to be aware of the storage location or system of the target files. The Search Agent conducts trial and error in formulating and revising search conditions, significantly saving time and effort until an answer is obtained.

Similar to user searches, the Search Agent can fully utilize QuickSolution®'s advanced search functions, such as aggregating similar documents, comprehensive score ranking*4, and hybrid search that combines keyword search and vector search*5. This enables highly accurate searches for information necessary for answering, without omissions or duplication, and allows precise answers to be obtained from generative AI.

Enhanced generative AI collaboration (RAG) function enables even more diverse usage

By equipping the Search Agent and enhancing the generative AI collaboration (RAG) function, QuickSolution® can now be used in various ways depending on the situation, such as:

• Ask a question to all search targets (when the storage location is unknown)
  Simply by asking a question in chat, the Search Agent autonomously searches and evaluates, generating precise answers.

• Specify search target/folder and ask a question (when the storage location is known)
  If the target range (e.g., a specific department's folder) is known, you can specify the range and ask a question, allowing you to reach the desired information more quickly.

• Ask a question from narrowed-down search results (high-precision search → answer)

  First, narrow down candidate documents with QuickSolution®'s high-precision search, then ask generative AI a question about those results to get more precise answers.

• Ask a question/give instructions for a specific document (when the target file is clear)
  You can ask questions about specific documents. You can also register routine instructions, such as summarizing or translating English materials, and execute them with a single click.

In this way, when the "storage location is unknown," the Search Agent can perform autonomous exploration, and when the "storage location is known," you can specify the target for efficiency, enabling optimal operation according to the situation.

Generative AI collaboration (RAG) that can be introduced without changing existing operations and used with confidence

QuickSolution®'s generative AI collaboration (RAG) can use internal search targets as information sources as they are, allowing it to be introduced without changing existing operations.

In addition, QuickSolution®'s high-precision search and aggregation of similar documents appropriately select necessary documents and provide precise answers. Furthermore, it generates answers considering the user's viewing permissions, and by operating in a way that internal corporate information is not used for generative AI training, it reduces the risk of confidential information leakage.

Through these features, QuickSolution® maximizes the potential value of internal information and further promotes operational efficiency, knowledge utilization, and DX.

In the enterprise search market of the AI agent era, where demand is expected to grow further, QuickSolution® will continue to develop products that anticipate the needs of the times, under the keyword "Reforming 'how to search' with AI!".

*1: RAG (Retrieval-Augmented Generation) is a technology that searches external information sources, extracts necessary parts, and generates answers from large language models (LLMs) based on that content. For details, please refer to [What is RAG (https://www.sei-info.co.jp/quicksolution/column/rag/?argument=EkZKmXCQ&dmai=20260407_PR)].

*2: QuickSolution®'s generative AI collaboration (RAG) function is optional. It supports OpenAI's ChatGPT, Microsoft's Azure OpenAI Service, Google's Gemini (Vertex AI), Anthropic's Claude, Amazon's Amazon Bedrock, and others. By collaborating with the latest generative AI models such as GPT-5.1, GPT-5.1 chat, Gemini 3.1 Pro, Claude Sonnet 4.5, and Claude Opus 4.5, internal information RAG can be realized. By using Azure OpenAI Service, generative AI can be utilized in a highly secure environment. Also, by using Amazon Bedrock in AWS's closed network environment, generative AI can be utilized securely.

*3: QuickSolution® achieved a quadruple crown, ranking No. 1 in the following four market research reports:

Fuji Chimera Research Institute "Software Business New Market 2017-2025 Edition" Search Engine Package Market (Vendor Share 2016-2024 FY) (9 consecutive years)

ITR "ITR Market View: Generative AI/Machine Learning Platform Market 2025" Search/Discovery Engine Market: Vendor-specific Sales Share (2017-2024 FY) (8 consecutive years)

Deloitte Tohmatsu MIC Research Institute "Collaboration & Mobile Management Software Market Outlook 2019-2025 FY" Enterprise Search/Search Engine Market: Vendor-specific Shipment Value/Number of Companies Share (2018-2024 FY) (7 consecutive years) (https://mic-r.co.jp/mr/03500/)

Fuji Chimera Research Institute "Comprehensive Survey of Artificial Intelligence Business 2022 Edition" Knowledge Management Solution Market (2021 FY)

*4: Comprehensive score ranking is a method of ranking search results by taking into account full-text scores, data freshness, number of accesses, and more.

*5: Vector search is a method of searching data by digitizing (vectorizing) the "meaning" of words and sentences and using their similarity. Unlike conventional keyword matching searches, a major feature is the ability to find "semantically similar information" even if the notation is different. For details, please refer to [What is Vector Search? Utilizing Hybrid Search (https://www.sei-info.co.jp/quicksolution/column/vector-search/?argument=EkZKmXCQ&dmai=20260407_PR)].

End

■ About QuickSolution®

・Website

https://www.sei-info.co.jp/quicksolution/?argument=EkZKmXCQ&dmai=20260407_PR

・QuickSolution® Catalog & Case Studies in 5 Minutes

In addition to simple explanatory materials for QuickSolution®, successful case studies of business efficiency improvement, DX, and knowledge management can be downloaded from the page below.

https://www.sei-info.co.jp/quicksolution/download/?argument=EkZKmXCQ&dmai=20260407_PR

・Webinar Information

We select one theme from highly adopted utilization methods and introduce how to use QuickSolution® through examples and demonstrations. Those interested in generative AI collaboration (RAG) are also welcome to join.

　Business Reform with Enterprise Search × Generative AI

　・Business efficiency improvement with enterprise search

　・Common challenges and solutions for knowledge utilization

　・The optimal solution for RAG, the effects of enterprise search × generative AI

　・Enterprise Search Agent ~The forefront of information retrieval~

Apply here

https://www.sei-info.co.jp/quicksolution/event/webinar-qs/?argument=EkZKmXCQ&dmai=20260407_PR

・Main Features

Large Capacity: Searches 100TB (100 million items) scale in 2-3 seconds on 1 server, expandable to several hundreds of TB, cross-searches a rich set of targets including on-premise/cloud (SharePoint Online, Box, Garoon, etc.)

High Accuracy: Displays important documents at the top without search omissions, prevents duplication of results with aggregated display of similar documents, semantic search (hybrid of string & vector search), relevance feedback

Thumbnails/Viewer: Thumbnail display in search results, direct display/highlight of hit locations in viewer

Generative AI collaboration: RAG compatible, generative AI answers internal information, no file upload required, highly accurate answers through search without omissions or duplication, considers viewing permissions

Easy and Secure Operation: Intuitive operability, knowledge transfer with Click Navi®, comprehensive permission management

Organization/Classification/Utilization: Organize/classify with favorites/shared tags, utilize information with Know-Who

Comprehensive Analysis Functions: Visualization of search results, document classification/clustering, thesaurus builder, named entity extraction

Extensive Track Record: Use from smartphones/tablets, operation in cloud and virtual environments, multi-language support

System Integration: Purely domestic, rich API, seamless integration with other systems

Diverse Search Targets: File servers (including Amazon FSx, PowerScale (formerly Isilon), NetApp, etc.), websites, databases, Notes, Microsoft 365, Box, Google Drive, Garoon, ArcSuite, Rakuraku Document Plus, Rakuraku Workflow II, Rakuraku Web Database, etc. cross-search

■ About Registered Trademarks

・QuickSolution, QuickSolution, Rakuraku Document, Rakuraku Workflow are registered trademarks of Sumitomo Electric Industries, Ltd.

・ChatGPT, GPT, OpenAI are trademarks or registered trademarks of OpenAI, Inc. or its subsidiaries in the United States and other countries.

・Google, Google Drive, Gemini, Vertex AI are trademarks or registered trademarks of Google LLC in the United States and other countries.

・Claude is a registered trademark of Anthropic PBC.

・Amazon Bedrock, AWS, Amazon FSx are trademarks or registered trademarks of Amazon.com, Inc. or its affiliates in the United States and other countries.

・Click Navi, Rakuraku Web Database are registered trademarks of Sumitomo Electric Information Systems Co., Ltd.

・PowerScale, Isilon are trademarks or registered trademarks of Dell Inc. or its subsidiaries in the United States and other countries.

・NetApp is a trademark or registered trademark of NetApp, Inc. in the United States and other countries.

・Notes is a trademark or registered trademark of HCL Technologies Ltd.

・Microsoft, Microsoft 365, Azure are trademarks or registered trademarks of Microsoft Corporation in the United States and other countries.

・Box is a trademark or registered trademark of Box Inc. in the United States and other countries.

・Garoon is a trademark or registered trademark of Cybozu, Inc.

・ArcSuite is a registered trademark of FUJIFILM Business Innovation Corp.

■ Overview of Sumitomo Electric Information Systems Co., Ltd.

・Established: October 1, 1998

・Capital: 480 million yen

・Number of employees: 670

・Representative: Koichi Kuratomi, President

・Head office location: Nissei Shin-Osaka Building, 3-4-30 Miyahara, Yodogawa-ku, Osaka

・Business content: Development, sales, and maintenance of various business software, development, sales, and maintenance of package software, provision of cloud services