top of page
Academic_640x160_en.png

Chapter 7: Research & Knowledge Management—How RAG and Enterprise Search Are Transforming “Research Work”

  • 21 hours ago
  • 6 min read

7-1. Why “Internal Search” and Knowledge Are Becoming Strategic Priorities

Across organizations worldwide, a common challenge persists: employees cannot easily find the information they need. Studies suggest that knowledge workers spend roughly 20–30% of their working time searching for information, resulting in productivity losses amounting to thousands of dollars per employee annually. Japanese companies face similar issues, including “Excel files known only to specific individuals,” “labyrinthine shared folders,” and “outdated internal wikis.”

Against this backdrop, the AI-driven knowledge management market is projected to grow from $5.23 billion in 2024 to $7.71 billion in 2025, representing a CAGR of approximately 47.2%. At the core of this growth are enterprise AI search and RAG (Retrieval-Augmented Generation)-based knowledge platforms.

Globally, investment trends show that 28% of enterprise generative AI spending is allocated to search and retrieval, and 27% to data extraction and transformation. This indicates that organizations are prioritizing the ability to effectively utilize internal information as a foundational step in AI adoption.



7-2. What Are RAG and Enterprise AI Search?

RAG (Retrieval-Augmented Generation) is an architecture that combines the generative capabilities of large language models (LLMs) with the retrieval capabilities of search systems. When a user asks a question, the system first retrieves relevant information from internal and external data sources, then uses that context to generate a response.

This approach reduces the risk of hallucinations—where AI generates unsupported or incorrect information—while enabling answers that synthesize data across internal policies, reports, meeting notes, emails, and support tickets.

Enterprise AI search platforms such as Glean connect to over 100 SaaS and internal systems, delivering real-time search, summarization, and answers based on user permissions. According to Forrester, such tools can reduce search time by up to 110 hours per user annually.

For Japanese companies, this type of permission-aware, context-rich search represents a new knowledge infrastructure that balances productivity gains with strict information security requirements.



7-3. Generative AI as a New Search Interface

Experts in knowledge management describe generative AI as the most significant shift in search interfaces in the past 30 years. Traditional search relied heavily on keyword matching, requiring users to know exactly what to search for.

With generative AI-powered systems, users can instead ask natural language questions such as:“What were the key pricing strategies of competitor A in North America last year?”“Summarize ESG-related risks identified in the past six months.”“What arguments were effective in similar past proposals for this client?”

The system retrieves relevant documents, synthesizes the information, and delivers a structured response. In this sense, generative AI eliminates the burden of crafting search queries and enables users to interact with information in the way they naturally think.



7-4. Market Signals: Knowledge AI as Infrastructure

Investment trends suggest that AI solutions in knowledge management are rapidly becoming infrastructure-level technologies. With a growth rate of approximately 47% between 2024 and 2025, this category is among the fastest-growing in enterprise IT.

Glean, a leading enterprise AI search company, raised $150 million in its Series F round in 2025, reaching a valuation of $7.2 billion. The company has surpassed $100 million in annual recurring revenue and is deployed across hundreds of organizations, including Fortune 500 companies.

Such investment levels indicate that building an “internal knowledge internet” is becoming as critical as cloud infrastructure or CRM systems.

At the same time, global surveys show that 36% of respondents identify weak technical infrastructure as a major barrier to effective knowledge management. Generative AI and enterprise search are increasingly seen as a way to overcome these structural limitations.



7-5. Expansion of AI for Research Use in Japan

According to a 2025 survey by GMO Research & AI, 31.2% of Japanese business professionals are already using generative AI in their work. Among them, 25.5% use it for research and information gathering. When combined with tasks such as report and slide creation (43.2%), writing and translation (51.5%), and idea generation (35.3%), it becomes clear that AI is embedded across the entire workflow of “researching, summarizing, and writing.”

PwC Japan’s 2025 survey similarly shows widespread use of generative AI for research, document preparation, and summarization. However, many companies still limit AI usage to publicly available information and non-sensitive internal data, avoiding integration with confidential systems.

As enterprise search and RAG platforms mature, this will shift from “AI that searches public information” to “AI that integrates internal and external knowledge,” fundamentally transforming research quality.



7-6. Knowledge Creation in Japan: The RIETI Perspective

The Research Institute of Economy, Trade and Industry (RIETI) highlights an important distinction: generative AI excels at recombining existing explicit knowledge but does not inherently create new knowledge.

According to RIETI:Explicit knowledge (documented information) can be effectively processed, summarized, and recombined by AI.Tacit knowledge (experience-based insights) is generated through human interaction and real-world practice.

This distinction has important implications for enterprise knowledge strategy. AI should be used to surface and structure existing knowledge, while humans remain responsible for generating new insights through experience and collaboration.

In this model, organizations must design knowledge systems where human-generated tacit knowledge is continuously converted into explicit knowledge that AI can utilize.



7-7. Practical Use Cases: Automating Research and Knowledge Work

A growing number of use cases are emerging globally and in Japan:

Automated market and competitor researchAI aggregates external data, news, and internal reports to generate structured summaries of market size, key players, and trends. Researchers then refine these outputs with insights and hypotheses.

Cross-organizational knowledge searchEnterprise search systems connect emails, chat logs, internal wikis, CRM notes, and ticket histories, enabling users to instantly find experts, past cases, and reusable templates.

Automated project briefingsAt project kickoff, AI compiles relevant past cases, client data, and technical documents into a concise briefing report that can be reviewed in under 30 minutes.

Automated knowledge captureMeeting transcripts and chat logs are automatically summarized, tagged, and stored in knowledge bases, converting tacit knowledge into searchable explicit knowledge.

These use cases dramatically reduce the time required for preparatory work—from days to minutes—allowing professionals to focus on higher-value activities such as hypothesis building, interviews, and field observation.



7-8. Challenges in Japan: Data Silos and Access Control

Despite its potential, implementing knowledge AI in Japan presents several challenges:Data is fragmented across departments (silos)Documents are unstructured and scattered across systemsAccess control policies are often unclear or inconsistently defined

Without proper governance, connecting all data sources to an LLM can introduce risks of information leakage and unauthorized access. In fact, 44.3% of Japanese users cite accuracy concerns, and 34.9% cite security concerns regarding generative AI.

To address these issues, enterprise AI search platforms enforce strict permission inheritance from source systems, ensuring that users only see data they are authorized to access. This design is particularly important in Japan, where access control and auditability are often more stringent than in other markets.



7-9. Organizational and Talent Implications

The adoption of knowledge AI is not just a technological shift—it requires changes in organizational culture and skills. PwC surveys indicate that the biggest bottleneck in generative AI adoption for Japanese companies is talent and skill development.

Key role shifts include:Researchers: from information gathering to question design and insight generationKnowledge managers: from document management to structuring knowledge for AI usabilityDomain experts: from tacit knowledge holders to contributors of reusable organizational knowledge

Japanese companies have traditionally relied on highly skilled individuals. In the AI era, competitive advantage will depend on how effectively these individual insights are captured and transformed into organizational knowledge.



7-10. Bridge to the Next Chapter: HR and Talent Management

The automation of research and knowledge management is transforming how organizations access and use information. At the same time, it is reshaping how companies understand, deploy, and develop talent.

If AI can map who knows what, which skills exist within the organization, and how individuals have contributed to past projects, it fundamentally changes HR and talent management.

In the next chapter, we will explore how AI is transforming recruitment, onboarding, skill mapping, performance evaluation, and workforce development—linking organizational knowledge with human potential in both global and Japanese contexts.



reference

Comments

Latest Articles
archive

© JASEC 2017

Japan E-Commerce Association

Japan Academic Society for E-Commerce

 

Shoji NISHIMURA Lab., Faculty of Human Sciences, Waseda Univ.
2-579-15 Mikajima, Tokorozawa, Saitama 359-1192, Japan

info@jasec.or.jp +81-4-2947-6717

  • meta-70x70
  • X
  • Youtube
  • JASEC 一般社団法人 日本イーコマース学会：LinkedIn
bottom of page