agent memory as graphs or traces

[davidberenstein1957]

Is your feature request related to a problem? Please describe.
Agents are cool but then can be a bit black-boxy. Logs and memory solve this a bit but they show a per-run overview, which requires more focus/attention to actually process as a human.

Describe the solution you'd like

• Convert them to traces which could be visualised in something like langfuse.
• Convert them to a graph.
  
• Convert them to process logs https://www.processmining.org/event-data.html. Might be used to indicate failure points in more complex agent flow, which I used for chatbot analysis a long time ago, although this is more researchy.

Is this not possible with the current options.
This is not an out-of-the-box integration.

Describe alternatives you've considered
Looking at the logs.

Additional context
N.A.

[davidberenstein1957]

somewhat related to #419 (comment)