Retrieval-augmented generation (RAG) is a tool that improves LLMs by retrieving sensitive information from files and databases. LLMs can use that data to answer questions outside their training data and generate more accurate and up-to-date responses.

GraphRAG is optimized RAG. It stores data within a graph-based data repository to make better conceptual relationships between words and sentences. Our benchmark results show that GraphRAG increases the accuracy of the response generated by up to 30%. GraphRAG is especially great at processing tasks with large data and can take longer to generate a response.

Autonomous agents are systems that can complete specific tasks without human intervention. By serving agents over APIs, you can easily add agents capabilities into any application to handle even the most complicated real-world business tasks. Agentic generative AI models can understand the environment and what users want to achieve and complete the required actions to fulfill them.

Optimized open-source LLM models allow you to create simple generative AI-powered assistants and private ChatGPT-like tools without connection to private data. It is the easiest and cheapest way to add AI features to your employees and customers.