Perplexity researchers published findings showing their Computer product performs 26 minutes of autonomous work per user session compared to 33 seconds for traditional search. The paper "How AI Agents Reshape Knowledge Work: Autonomy, Efficiency, and Scope" by Jeremy Yang, Kate Zyskowski, Noah Yonack, and Jerry Ma analyzed production data from Perplexity's Search and Computer products to measure how AI agents transform knowledge work.
Computer Reduces Task Completion Time by 87 Percent
The research identified three key findings using near-identical initial query pairs as natural experiments for the same underlying task. Computer automates task decomposition and execution that Search users manually orchestrate, shifting follow-up query distribution toward higher-order work like verification and extension. Per-query dissatisfaction rates are 55 percent lower on Computer than Search.
Computer reduces completion time from 269 to 36 minutes on matched tasks, lowering estimated time and cost by 87 percent and 94 percent respectively compared to humans equipped with Search alone. The efficiency gains stem from autonomous agents handling multi-step processes without requiring user intervention between steps.
AI Agents Expand Scope of Attempted Work
The study found Computer fundamentally changes what users attempt. Queries more often cross occupational boundaries, require higher-order cognition, draw on broader expertise, take the form of composite tasks bundling interdependent subtasks into a single query, and unlock work activities essentially absent from Search usage among the same users.
The methodology controlled for task difficulty and user intent by comparing near-identical queries. The core implication: "AI agents accelerate workflows, enhance output quality, reduce costs, and expand the breadth and depth of automated work." The evidence demonstrates autonomous agents don't merely accelerate existing workflows—they fundamentally reshape what knowledge work users attempt and how they approach complex problems. The paper was published June 5, 2026 on arXiv (2606.07489).
Key Takeaways
- Perplexity Computer performs 26 minutes of autonomous work per session versus 33 seconds for traditional search
- Task completion time reduced from 269 to 36 minutes, cutting time and cost by 87 percent and 94 percent
- Per-query dissatisfaction rates are 55 percent lower on Computer compared to Search
- AI agents enable users to attempt work crossing occupational boundaries and requiring higher-order cognition
- Research published June 5, 2026 on arXiv using production data from matched query pairs