We're sharing a comprehensive study of Perplexity Computer in real-world deployment in collaboration with Harvard

Computer is more cost and time-efficient, unlocks cross-disciplinary search beyond the reach of multi-step search, and provides higher autonomy and quality in the generated outputs.

Search is cheap to start but costly per step, since the human runs each one.

Computer costs more upfront to delegate and verify, but little per step, since the agent runs the steps.

It pays off once a task is long enough to clear that fixed cost, expanding the affordable frontier of work worth attempting.

These results show that Computer increases autonomy, improves quality, cuts time and cost, and expands the scope of tasks users can attempt.

Read the full paper: https://arxiv.org/abs/2606.07489

We found that more autonomy with autonomous agents like Computer tracks with higher quality and satisfaction.

To compare the two products, we built 10,000 matched pairs where the same user asked near-identical queries to both Search and Computer, and Computer ran at least one execution tool.

We then priced human time using BLS wages.

Cumulative Computer queries reached 84x their first-week level by the end of the study. Most of them knowledge work queries.

Research and analysis (25.8%) and document and asset creation (18.6%) lead, concentrated in software, finance, and marketing.

Computer queries were nearly 3x as likely as Search queries to require expertise across three or more fields, like law, finance, or biology, in a single request.

23% involved work that never showed up in the same users' Search history.

Per session, Computer runs about 26 minutes of machine work to Search's 33 seconds.

It also reaches outside the harness far more, where 7.9% of Computer sessions call an external connector (avg 15 calls per session) versus 1.8% for Search.

@perplexity_ai @grok what does this mean to the layman employed in a normal company somewhere in a remote place

@AravSrinivas was just going through this research we r moving from AI as an assistant to AI as an operator. Research like this helps quantify that shift

@perplexity_ai This tracks with the shift we're watching—from AI as tool to AI as teammate. That's the real differentiator emerging. Autonomy builds trust, and trust builds adoption.

@perplexity_ai The 23% stat reveals massive latent demand for complex queries traditional search can't handle. This shift is where AI search becomes essential—bridging expertise gaps across domains.

@perplexity_ai This paper validates the shift we're tracking in SaaS—autonomous agents aren't theoretical, they're delivering 30-50% productivity gains in real workflows.

@perplexity_ai Big shift ahead: from searching for answers to agents getting the work done. 🚀🤖

@perplexity_ai 94% cost reduction is a real number if the baseline is paying someone to do search-heavy research manually. would want to see the task taxonomy before calling it generalizable

@perplexity_ai the 94% cost reduction is the number that will get screenshotted in every board deck - but the question it quietly opens is: cost to whom, and captured by whom

@perplexity_ai Strong evidence that autonomous agents are moving beyond chat. 87% faster, 94% cheaper, and higher user satisfaction is a compelling result.

@AravSrinivas the issue with these real-world deployment studies is they measure efficiency not whether anyone actually wants to use it for more than 3 days straight

@AravSrinivas State of the art orchestration! Perplexity always exceeds expectations 👍👍👍

@perplexity_ai This is where agents become interesting: not just answering questions, but actually reducing the friction around knowledge work. The key test is whether the time savings hold up outside controlled demos.