Agentcloud

Agent Cloud is an open source platform for building chat apps and agents that talk to a company's own data through retrieval augmented generation. Its core is a built in data pipeline that syncs, splits, chunks, embeds and stores data from a large catalogue of sources in a vector database, then keeps that data fresh on a manual, scheduled or cron based cadence so agents answer from current information. It is aimed at developers and internal platform teams that want private RAG applications. Users pick a connector such as Confluence or Google Drive, upload PDF, DOCX, TXT or CSV files, or select specific tables and columns from databases including Postgres, Snowflake and BigQuery. Embedding can use OpenAI's text embedding models or open source models such as BGE. The platform is LLM agnostic: bring an OpenAI key, use one of the cloud hosted open models, or connect a locally hosted model for a fully private setup. Agents can be given custom tools that call third party APIs, teams of agents can collaborate, and access controls are designed to stop chat apps from sharing more data than they should. Self hosted deployments connect to major vector databases such as Qdrant and Pinecone, while the cloud version includes a built in Qdrant store and APIs. No pricing page is published. The Community edition is free under the AGPL 3.0 copyleft license and runs locally via Docker, where the vendor recommends a machine with at least 16 GB of RAM. Enterprise and managed cloud terms are handled by contacting the team. Because the license is AGPL 3.0, organizations that modify the code and offer it as a network service should review their obligations to publish those changes.