agent-cma

To extract durable memories from conversation logs, you can use LLMs to parse the data and automatically identify user preferences, decisions, blockers, and goals, storing them in a structured format for long-term recall.

Yes, you can use Ollama or OpenAI models for conversation analysis to process chat logs, identify recurring issues, and enrich your knowledge base with structured durable memories.

The best way to automate extracting user goals and blockers from chat data is by applying LLM processing to daily conversation exports, which parses the logs and consolidates identified memory artifacts into a knowledge base.

Yes, extracting durable memories requires access to conversation logs for processing and LLM services, such as Ollama or OpenAI, to analyze the data and store the extracted artifacts.

LLM-based memory extraction works by parsing customer support chat logs to automatically identify recurring issues as blockers, feature requests as goals, and expressed likes or dislikes as preferences, creating a consolidated knowledge base.

Limitations of using LLMs for conversation data processing include the dependency on external LLM services for analysis and the requirement of having properly formatted conversation exports available to parse for durable memories.