RAGent is a retrieval-augmented generation (RAG) platform developed by SRG, the cross-organizational site reliability engineering group at CyberAgent. It turns a range of workplace tools — Kibela, Slack and spreadsheets among them — into data sources that can be searched by vector search, so that past documents and related knowledge can be looked up efficiently.
Capabilities
RAGent indexes content from tools such as Kibela, Slack and spreadsheets for vector search and is reachable by an AI agent through a Model Context Protocol server (RAGent MCP).
Adoption & Ecosystem
At WINTICKET it serves as the search layer for AI-assisted specification writing: Devin queries it through RAGent MCP to find similar past specifications and to consult writing guidelines kept in Kibela, alongside DeepWiki for reading code, as described in WINTICKETで仕様書作成にAIが定着するまでにやったこと.