Indexify is a knowledge and memory retrieval service for Large Language Models. It facilitates in-context learning of LLMs by providing relevant context in a prompt or exposing relevant memory to AI agents.
Indexify has a built in data parallel content extraction engine which enables scalable extraciton of content using AI models to keep indexes updated in real time(or re-index with a new model) as data is ingested.
- Knowledge Base for LLMs: Real time retrieval of knowledge and context from private documents and structured data to improve accuracy of LLM models.
- Memory Engine for Co-Pilot agents: Store and retrieve long-term memory of agents in real-time, providing enhanced personalization and improved user experiences for co-pilot and chat based applications.
- Real Time Data: Data connectors keep indexes updated by syncing with source automatically wherever possible, such as a S3 bucket or a database. This enables LLMs to answer queries that require real time information about the world.
- Secure Access: Apply role-based access control (RBAC) to ensure your sensitive data is exposed only to the applications you choose.
- State of the Art Embedding Models: Support for OpenAI and state of the art embedding models from HuggingFace.
- Custom Embedding Models: Import and use your custom-trained or fine-tuned HuggingFace Transformer-based embedding models with ease.
To get started follow our documentation.
Our comprehensive documentation is available - https://getindexify.ai
Please open an issue to discuss new features, or join our Discord group. Contributions are welcome, there are a bunch of open tasks we could use help with!
Join the Discord Server - https://discord.gg/mrXrq3DmV8