AI Meeting Minutes Tool 'Rimo Voice' Launches 'AI Search' for Cross-Document Conversational Retrieval

Rimo Inc. (Headquarters: Shibuya, Tokyo; CEO: Naoyoshi Aikawa) has officially released the 'AI Search' feature for its AI meeting minutes tool, 'Rimo Voice,' enabling cross-document searches of accumulated meeting and business negotiation records via a chat interface. With 'AI Search,' users can instantly obtain answers with citations by simply asking questions in a chat, limited to records they have permission to view. The feature has been validated using approximately 600 million characters of proprietary data accumulated over the six years since Rimo Voice's launch, achieving high-precision cross-search capabilities powered by AI specialized in the Japanese language. This transforms meeting and negotiation records that often get buried within a company into valuable information assets for future decision-making. As the volume of records grows, the time spent searching for specific information increases. It is estimated that business professionals spend an average of 1 to 2 hours per day searching for information, and records becoming 'dormant assets' remains a challenge even in the AI era. 'AI Search' allows users to extract relevant information from transcripts and minutes accumulated in Rimo Voice instantly. There is no need to read through records one by one. Whether personnel changes or projects span across departments, records can be utilized as organizational assets. Answers include links to the source meeting records, allowing users to verify the basis of the information immediately. Furthermore, search targets are limited to notes the user has permission to view, eliminating concerns about unauthorized access to confidential information or records from other departments. Rimo Voice's 'AI Search' is designed with three key features to ensure information assets are used safely and efficiently in daily operations: 1. Immediate access to source meetings. 2. Maintenance of internal information management rules. 3. High-precision analysis of Japanese meeting minutes. Generic RAG implementations often struggle with the nuances of Japanese, such as variations in expression, colloquialisms, and technical jargon. Rimo Voice's 'AI Search' utilizes proprietary fine-tuning based on vast amounts of Japanese audio data accumulated since the service's inception. By continuously optimizing for colloquialisms, jargon, and linguistic variations, it achieves answer accuracy specialized for the Japanese meeting minutes domain. Additionally, the company conducted operational verification using over 30,000 hours and 600 million characters of data, confirming stable performance even with large-scale accumulated data. To use the feature, simply click the 'AI Search' button next to the search bar in the header. Looking ahead, while current searches are limited to Rimo Voice records, the company plans to expand the scope to include internal manuals and external documents, aiming to create an environment where various organizational information assets can be utilized through AI search. As of March 2026, 'Rimo Voice' is used by over 250,000 accounts in Japan and has won 19 awards at the BOXIL SaaS AWARD.