Ollmo, a production-grade RAG and Agent visualization platform for private deployment and team collaboration, has officially released version 0.1.0. This platform aims to help users quickly obtain answers with precise sources by uploading documents and asking questions, and to easily set up a private knowledge base Q&A system out of the box.

In terms of core technology and architecture, Ollmo adopts a front-end and back-end separation design. The backend is built using Go and Fiber, while the frontend relies on Next.js App Router and shadcn-ui. It fully supports Docker one-click deployment and offers two service models: private deployment and SaaS. Its built-in hybrid search engine integrates Milvus dense vectors and MySQL full-text search, supporting RRF score fusion and visual adjustment of keyword and semantic weights; multi-tenant isolation is achieved through full-chain logical isolation of MySQL, Milvus, and MinIO.

Additionally, the platform is equipped with a rich set of intelligent and engineering tools. Asynchronous document pipelines support batch uploads and intelligent chunking; streaming溯源对话 can output answers in real time with document citations; manual annotation of knowledge base Q&As and automatic session memory summary functions effectively reduce system hallucinations and maintain context; the Agent visualization canvas built with React-Flow supports flexible workflow orchestration and single-node debugging; GraphRAG technology can automatically extract entities to build knowledge graphs; meanwhile, the system integrates usage statistics and cost dashboards, full-chain observability tracing, team permission management, a management console, open APIs, and bilingual internationalization support, comprehensively meeting the development and collaboration needs of enterprise-level teams.