Perplexity Launches Brain, a Self-Improving Memory System That Builds a Context Graph of an Agent’s Work and Learns Overnight
Perplexity's Computer agent just got a memory upgrade. Brain learns from its own mistakes overnight—improving correctness, recall, and cost without human intervention.
Why it matters
Perplexity is shipping a self-improving agent feature that learns from execution traces, not user feedback. This is a concrete example of agentic systems becoming self-optimizing—a capability shift that changes how founders should think about agent reliability and cost curves.
The key facts
6 to knowPerplexity launches Brain—a memory system for Computer agent
Brain tracks agent work: successes, failures, corrections
Builds traceable context graph of agent operations
Self-improves overnight via graph review
Early gains reported in: correctness, recall, cost efficiency
Published June 18, 2026
Go to the source
MarkTechPostmarktechpost.com
Publisher excerpt: Perplexity has launched Brain, a self-improving memory system for its Computer agent. Instead of remembering the user, Brain remembers the agent's work — what worked, what failed, and what corrections got made. It builds a traceable context graph, reviews it overnight, and reports early gains in…