Perplexity Launches Brain, a Self-Improving Memory System That Builds a Context Graph of an Agent’s Work and Learns Overnight

Perplexity has introduced Brain, a self-improving memory system designed for its Computer agent, which focuses on tracking the agent's work rather than user interactions. This system constructs a context graph that captures the agent's performance, including successes and failures, and utilizes overnight reviews to enhance correctness, recall, and cost-effectiveness. This development is significant for practitioners as it enables agents to learn from their operational history, potentially leading to more efficient and intelligent AI systems.