The problem with using context in AI Agent + Postgres Chat Memory: the agent stops calling tools after 2-3 requests

Good day everyone!
I came across an unobvious behavior in workflow, and I hope that someone will be able to suggest a solution.

Configuration:

  • Workflow implements chat with a PostgreSQL table via text-to-SQL.

  • All control is delegated to the AI Agent, which chooses which tool to use.

  • Postgres nodes are used as tools (they execute SQL queries).

  • To keep the context between requests, I connected the Postgres Chat Memory node (the memory depth is the last 10 interactions).

Problem:

  • Everything works correctly without memory: the agent calls Postgres tools every time and returns accurate results.

  • With the memory enabled, the agent does “remember” the previous questions, but only during the first two clarifications.

  • Starting with the third request in the chain, the agent stops invoking tools and instead generates fictional responses (hallucinations).

Question:
What could be the reason for this behavior?

  • Perhaps the problem is context overflow (although the depth is only 10)?

  • Or does the agent “lose confidence” in the need to call tools as the length of the dialog increases?

  • Are there any recommendations for setting up memory or prompta for such scenarios?

Any ideas or links to the documentation are welcome!

1 Like

IT’S DECIDED!:partying_face::tada:
A deep investigation of the issue using Perplexity (not advertising) helped, he recommended installing the n8n-nodes-better-ai-agent community node, the node itself is searched for by the name Better AI Agent.
Now all 10 iterations: the question-answer model accesses memory (it works with both Simple Memory and Redis and Postgres, in general, the problem is not in memory) and calls tool_calls (the problem is not in DeepSeek).

To summarize, it turns out that this was a node bug and the community fixed it before the developers (hopefully, everything will be fixed in version 2.0).

1 Like

did you find any solution in in native n8n AI Agent