- Improved chat experience with a memory layer: For this, we are using Mem0. This can be configured when creating the agent.
- Notion-style editing to configure agent: A sleeker UI and a familiar editing experience for our users.
- Encrypted API keys and only decrypted at the point of usage. We are using Infisical to store our encryption keys, ensuring greater security.
- Add knowledge bases to chat assistants. We are using Unstructured to handle the parsing of various document types and chunking for greater context. This is much better than our v1 where users need to add all content in the prompt.
Our goal is to build an open-source enterprise-grade AI platform that combines the best-in-class LLM tools (preferably open-source too) for users to build AI agents easily.
Thank you!
