Greetings.
Just getting started with n8n. Trying to put together an intent-based routing workflow, which will make use of AI Agent node to perform user intent classification. One of the requirements for this workflow is to be able to detect a follow-up requests, for which a long-term memory is required for the AI Agent node. I have tried using Postgres memory, but it seems unreliable and requires a fair amout of manual handling to insert and read memories. From my previous work on Google ADK-based agents, Mem0 was a great memory choice. Does anyone have any examples of using Mem0 as a long-term memory for their AI Agent nodes? Not as an MCP, but directly within the workflow.
Thanks!
It’s great you’re exploring the capabilities of n8n for intent-based routing and AI agent workflows! Your observation about Postgres memory and its reliability/manual handling aligns with some common discussions in the n8n community regarding long-term memory for AI agents.
In the n8n Community forum, there have been feature requests for native Mem0 support inside n8n, but as of now, there is no official Mem0 node available in n8n. Developers commented that Mem0 would be a major leap compared to existing memory options like Postgres or Redis. n8n Community, n8n Community
If you’d like to see this feature (like native Mem0 integration) become part of n8n, be sure to click the vote button at the top of these threads!
The n8n team actively monitors community interest, and your vote helps influence what gets prioritized and built into the product. Every vote counts.
