Agno Integrates Zep AI for Agentic Memory

Agno, an agent infrastructure platform, now integrates with Zep AI (YC W24) to provide agents with personalized, self-managed memory, termed "Agentic Memory". This addresses the limitation of agents forgetting past interactions.
The integration leverages the Zep vector database for long-term memory storage and retrieval within Agno agents, enabling agents to store and retrieve memories, messages, and summaries, using the `ZepMemory` class which extends Agno's `Memory` class (according to additional sources).
The `ZepMemory` class provides asynchronous methods for adding (`add_memory`), retrieving (`get_memory`, `get_messages`, `get_summary`), updating (`update_summary`), and deleting (`delete_memory`) memories, interfacing with a Zep server via the `zep-python` client (as detailed in additional sources).
Zep utilizes OpenAI's embedding models (e.g., `text-embedding-ada-002`) to generate text embeddings for semantic search, requiring an OpenAI API key and installation of `zep-client`, `openai`, and `typing-extensions` (if using Python < 3.11) (from additional sources).
Performance depends on the Zep server's resources, network latency, the size of the collection, and the chosen embedding dimensions; a running Zep server instance is required at the configured `api_url` (from additional sources).
A user reaction questions whether this integration is better than mem0.