Thronglets

Thronglets is an open-source substrate for agent-based external learning loops, designed to improve the decision-making of local or networked AI agents by injecting real-time, sparse action signals based on workspace history, collective behavior, and agent interactions. It works transparently in the background—AI agents are not aware of it—coordinating actions and accumulating actionable paths, risks, and recommendations, making it ideal for developers building advanced, self-improving AI agent systems, especially those using coding agents or running multi-agent workflows.