Agent with decisions memory at its core (github.com)

A new AI framework called Nous has been announced, designed to build intelligent agents capable of thinking, learning, and evolving over time, based on principles from Marvin Minsky's Society of Mind. Unlike traditional stateless AI agents that simply respond to prompts, Nous integrates a decision memory core, known as Cognition Engines, which allows agents to maintain structured memory and learn from past actions. This framework emphasizes self-monitoring and administrative growth, meaning agents can improve their cognitive processes instead of merely accumulating data. The significance of Nous lies in its potential to transform how AI agents function. By implementing features such as structured memory, decision intelligence, and explicit frame-switching, Nous aims to create agents that mimic human-like decision-making processes. Each agent follows a cycle of sensing inputs, selecting cognitive frames, deliberating, and monitoring outcomes, enabling a sophisticated learning loop. Ultimately, Nous targets a higher level of cognitive capability, moving AI agents from reactive, low-level operations to proactive, self-improving entities using advanced principles of cognitive science.