Details
- Perplexity introduced Brain, a continuously learning memory system that powers its Computer agent environment.
- Brain builds a persistent context graph so every Computer task can reuse knowledge from previous runs instead of starting from scratch.
- The system makes Computer more stateful over time, capturing projects, decisions, and sources across sessions.
- In internal evaluations on tasks requiring historical context, Brain improves answer correctness by 25% and recall by 16% while running 13% cheaper per task.
- Each stored memory links back to its original session, file, or source, giving users transparent traceability and control over what is retained.
- Brain is available as a research preview for all Perplexity Max subscribers and can be accessed under the Customize section in the app sidebar.
- Perplexity is publishing additional technical and product details about Brain through its official documentation and blog.
Impact
By adding a structured, persistent memory layer to its Computer environment, Perplexity moves closer to project-centric AI workflows that span many sessions, narrowing the gap with agents that promise long-lived context. The reported quality and cost gains on context-heavy tasks suggest that memory systems could become a key competitive dimension among frontier AI assistants.
