Brain stores agent memory as a linked Markdown wiki
- Source
- x.com
- Date
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.
Context
Perplexity's Brain gives Computer a way to consult accumulated knowledge without inserting the entire work history into every request. According to its architecture thread, Brain organizes sessions, files and sources into linked Markdown pages, a text-based wiki. Claims connect back to their supporting material, letting an follow a remembered statement to its evidence.
Separate background agents, called Dream agents, update that knowledge between active tasks. The design separates maintaining memory from answering a current request: the working agent retrieves relevant , while background processing reconciles new information with the stored account.
- 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.
- context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
- guardrails — The checks around a model that block bad inputs and outputs — filters, validators, and permission rules the model itself can't override.
- 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.
Checking sign-in…
Loading comments…