Brain is our self-improving memory system for Perplexity Computer. It compiles sessions, files, and sources into a structured knowledge wiki.
New evals build on our initial results, improving correctness by 9.3 points, currentness by 8.0, and recall by 8.9 with 15% fewer tokens.
Brain is a comprehensive memory system with three key components: durable memory storage, foreground agents that answer queries, and background agents that update and improve memory.
Brain stores memory as a filesystem of linked Markdown files.
It connects related subjects across the wiki and links each claim to the sessions and files that support it.
This structure allows us to organize memory perfectly suited for agents.
We use background agents, called Dream agents, to create a continuous self-improvement loop.
Dream agents run offline, with carefully defined scope and Dream-specific guardrails, synthesizing new information into consistent and accurate Brain updates.
guardrails — The checks around a model that block bad inputs and outputs — filters, validators, and permission rules the model itself can't override.
eval — A repeatable test for AI quality — a set of tasks plus scoring — used the way software teams use test suites, because model output is too variable to judge by eyeballing.
token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
Why it matters
A retrievable memory filesystem with background revision lets an AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → use a full history without loading it into every prompt, and the evalA repeatable test for AI quality — a set of tasks plus scoring — used the way software teams use test suites, because model output is too variable to judge by eyeballing.Full definition → show accuracy and tokenThe chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.Full definition → cost improving together.
Read the full thread on X ↗
repoI am getting genuine questionsbhanuhai2
articleFilesystem-Based Memory for LLM Agents: Organization, Evolution, and SustainabilitySizhe Zhou, Sheldon Yu, Hui Wei, Junda Wu, Siru Ouyang, Yizhu Jiao, Shijia Pan, Julian McAuley, Yu Zhang, Tong Yu, Jiawei Han