agentware-demo

Deploy an AI agent library that builds a dynamic knowledge base on the fly, enabling autonomous, context-aware conversations.

This Agentware demo showcases an AI agent library designed for persistent intelligence: it actively learns and updates its knowledge base during conversation. The system employs a robust client/server architecture (Source 1.4): the client manages LLM execution and memory, while the server utilizes a combination of vector and key-value databases for knowledge storage. For a clear example, the `examples/fish_location.py` script illustrates an agent, a housework robot, dynamically compressing memory with reflection. This process ensures unlimited conversation length without losing critical context or forgetting established facts, delivering a new standard for adaptive, long-term conversational AI.

https://github.com/agentware/agentware