Perplexity Enterprise Agentic-Specific Terms

Last Updated: 3 August 2026

These Agentic-Specific Terms (“Agentic Terms”) are supplemental to and governed by the Perplexity Enterprise Terms and Conditions (the “Enterprise Terms”). Except as expressly modified herein, all terms and conditions of the Enterprise Terms apply to Customer’s use of Perplexity’s services.

The provisions below apply solely to Customer’s use of Perplexity’s autonomous agentic features and products, including without limitation the Computer functionality and any other present or future agentic capabilities (collectively, the “Agentic Enterprise Service” or “Agentic Features”), and do not govern any other Perplexity services or features. To the extent of any conflict or inconsistency between these Agentic Terms and the Enterprise Terms, these Agentic Terms control only with respect to Customer’s use of the Agentic Features.

AUTONOMOUS AGENTIC FUNCTIONALITY IN ALL SERVICES.

(a) Applicability. The following provisions of this Section apply solely to the extent Customer uses the Agentic Features (as defined below), and solely to Customer’s use of the Agentic Features.

(b) Actions. Customer may use the Agentic Features to direct, initiate, or manage specific tasks or operations on behalf of Customer (“Actions”) that Customer may create, orchestrate, or initiate by submitting Inputs. For clarity, Actions may result in Output, but Actions are not themselves Output (for example, if Customer directs the Agentic Feature to create a research report, the research and other steps taken to create the report are Actions and the report itself is Output).

(c) Customer Responsibilities. Customer is responsible for (i) Actions taken by Agentic Features; (ii) determining whether and how Agentic Features should be used for Customer’s intended purpose; (iii) managing and controlling Agentic Features’ access to any data, applications, or systems (e.g., by logging into or out of third-party accounts such as email); (iv) exercising appropriate oversight to avoid undesired Actions; and (v) reviewing and approving Actions proposed by Agentic Features as needed. For clarity, all Actions taken by Agentic Features are deemed to be taken by the applicable Authorised User and not by Perplexity. Customer accepts full responsibility for such Actions.

(d) Security. Due to the nature of autonomous agentic artificial intelligence and the Agentic Features’ ability to browse the internet and interact with third-party systems, the Actions may be susceptible to security threats including but not limited to prompt injection attacks. Customer should supervise Agentic Features in order to stop any undesired or harmful Actions, and Customer (not Perplexity) is solely responsible for any Actions taken as a result of prompt injections or other attacks on the Agentic Features that could reasonably have been prevented by active Customer supervision of the Action while it was in progress.