How AI Agents Reshape Knowledge Work: Autonomy, Efficiency, and Scope

A study of Perplexity's autonomous AI agents reveals they perform 26 minutes of productive work per session versus 33 seconds for traditional search, reducing task completion time by 87% while improving quality and expanding the scope of work users attempt. This research demonstrates how AI agents are transitioning from conversational tools to end-to-end task executors that fundamentally reshape knowledge work.

The research presented here marks a critical inflection point in how AI systems create tangible economic value. Moving beyond conversational interfaces, autonomous agents are proving capable of orchestrating complex workflows without constant human intervention—a shift from reactive assistance to proactive problem-solving. Perplexity's production data provides empirical evidence that this autonomy advantage compounds across multiple dimensions: faster execution, higher quality outputs, and reduced cognitive load on users.

This transition reflects a broader maturation in AI capabilities. Earlier chatbot implementations required users to decompose problems manually and iterate through multiple prompts. Autonomous agents internalize this decomposition logic, handling subtask sequencing and decision-making independently. The 55% reduction in per-query dissatisfaction rates suggests users perceive higher-quality results, likely because agents can verify outputs and refine approaches without explicit prompting.

The economic implications are substantial. Reducing task completion time from 269 to 36 minutes translates to measurable productivity gains across knowledge-intensive sectors. The 94% cost reduction compared to human-only workflows signals potential displacement in routine analytical work while simultaneously enabling new use cases—the data shows users attempting more complex, cross-functional tasks on the agent platform.

Market participants should monitor adoption rates and industry applications closely. If these findings generalize beyond Perplexity's platform, demand for AI agent infrastructure accelerates significantly. The expansion of task scope suggests AI agents may unlock entirely new work categories rather than simply automating existing ones, creating opportunities for complementary tools and services while challenging traditional consulting and research workflows.

• →AI agents execute 26 minutes of autonomous work per session versus 33 seconds for traditional search, demonstrating substantial efficiency gains.
• →Task completion time and costs decrease by 87% and 94% respectively, with 55% lower dissatisfaction rates indicating quality improvements.
• →Autonomous agents shift user behavior toward higher-order tasks including verification, extension, and cross-occupational work previously uncommon in search-only workflows.
• →The autonomy advantage enables composite tasks bundling multiple subtasks, fundamentally changing the scope and nature of work attempted by users.
• →Production data from Perplexity provides empirical evidence that AI agents transition from conversational assistants to end-to-end task executors reshaping knowledge work.