Perplexity continued to sharpen its AI assistant platform this week, spotlighting a new “Brain” capability within its Computer product. Brain is described as a continuously learning memory system that builds a context graph across tasks, allowing workflows to become increasingly stateful over time.
Rather than starting from scratch on each query, Brain lets tasks begin with prior context from projects, decisions and sources. Early internal metrics suggest this approach can improve answer correctness by 25%, boost recall by 16% and cut per-task costs by 13% for context-dependent work.
The company emphasized transparency and user control by linking every memory back to its originating session, file or source. Brain is currently available as a research preview for Perplexity Max subscribers and can be enabled via the “Customize” section in the Computer sidebar.
This feature extends Perplexity’s broader push into autonomous AI agents and enterprise research tooling. Recent upgrades have tied its Deep Research capability directly into Computer, leveraging a “Search as Code” architecture that orchestrates large-scale retrieval and data processing for complex knowledge work.
If the reported efficiency gains from Brain and the wider Computer platform scale in broader use, Perplexity could deepen its appeal among professional and enterprise users who rely on persistent context and long-running workflows. These enhancements may support higher engagement, better unit economics and stronger competitive positioning in AI research and productivity markets.
Overall, the week underscored Perplexity’s strategy of differentiating through context-aware automation, measurable performance improvements and features aimed at intensive knowledge work use cases.