Cognita
A self-hosted RAG framework answers questions from documents using Ollama and local embeddings; it's Apache 2.0 licensed, archived and no longer maintained.
Cognita is a self-hosted RAG framework for developers building applications that answer questions from their own documents. The project is archived and no longer maintained.
- Document answers with references: Return answers alongside relevant document passages, with a browser interface for users to ask questions.
- Replaceable components: Choose parsing, embedding, retrieval and answer-generation components in a framework built on LangChain and LlamaIndex, and expose the application through an API.
- Incremental indexing: Process added or changed documents while skipping unchanged files, with support for local files, S3 buckets, databases and TrueFoundry Artifacts.
- Retrieval and agents: Use document retrieval for direct answers or build agents that perform multiple steps and use tools before responding.
The application runs locally through Docker Compose, including its web interface, metadata database and Qdrant vector storage. Local providers are enabled by default: Ollama runs LLMs, and Infinity supplies embeddings and reranking. SingleStore is another supported vector database.
OpenAI and TrueFoundry's model gateway are optional alternatives to local model providers; model requests go to those services when selected. Cognita is open source under the Apache 2.0 license.