I have been experimenting for a while now using Agents in LangFlow in a local Ollama based setup. I use LangFlow because my Python is not the best.
I have been very interested in the new neo4j-agent-memory because I think it will add massive benefit with the transparency of Agents in critical environments. I work with satellites and explainability is key.
Can anyone give some pointers on the best way to integrate neo4j-agent-memory into my LangFlow Agent? Any help would be greatly appreciated.
Hey @matt.kirk.smith - Can you say more about what kind of data your agents are working with? Is it like working with the raster data from satellite sensors? Or more about the satellites themselves?
I think this sounds like a good fit for the reasoning memory layer in neo4j-agent-memory.
I'm not very familar with LangFlow but it looks like since it is built on LangChain under the hood, the best integration path might be through neo4j-agent-memory's LangChain integration.
LangFlow seems to have a "custom component" feature that lets you write a Python class that wraps the memory client. I asked my intern to write up a complete snippet, I haven't tested it but might be a good place to start?