Written by

Perplexity Team

Published on

Nov 14, 2025

Comet Assistant puts you in control

Almost immediately after introducing Comet to the world, it was clear the Comet Assistant would set the standard for AI assistants, or agentic AI, on the web.  The Comet Assistant was an immediately popular tool that surprised even us. Today we’re announcing some upgrades to Comet Assistant.

As the world increasingly looks to Perplexity for leadership in agentic AI, it might be useful to outline our approach to building AI assistants.  

It’s the same approach for every assistant we build, whether it is the email assistant, background assistants, the new Comet Assistant, or another we’re excited to launch in the coming days. Your AI assistant is your AI assistant. It’s not the AI assistant, nor Perplexity’s AI assistant. It belongs to you, and we built it to power your curiosity and assist with anything you ask.

For this to work, our approach to AI assistants is grounded in three core principles: transparency in every action, user control over the assistant’s approach to assignments, and sound judgment by the assistant, knowing when to solicit feedback before taking action. All of this must take place on the bedrock of accurate AI that Perplexity builds to power answers for both human and agentic decision-making.  

Every improvement to the Comet Assistant advances these principles. It now shows you exactly what it’s doing, asks how you want it to act, and demonstrates sound judgment before completing sensitive actions. 

Transparency: Watch Comet Assistant work 

Trustworthy assistants show their work. Now it’s easier to see when the Comet Assistant is browsing for you and what actions it’s taking on your behalf. You can watch while Comet works or see its reasoning step-by-step in the Assistant sidecar. With the newest Comet Assistant, you’ll see exactly where it’s clicking, scrolling, or interacting with a site. And there are clear buttons to stop the Assistant or provide more guidance.

User Control: Decide when Comet Assistant acts 

AI assistants, like human assistants, must be trustworthy. Now, when you ask a question in the omnibar or Perplexity Search, Comet may ask if you want your Assistant to handle the task itself. You can choose to browse yourself, give Comet permission to browse on your behalf once, or let the Assistant automatically browse for you when it detects the chance to be useful. 

Sound judgment: Built-in safeguards for high-stakes decisions 

A great assistant knows some decisions require your input. When Comet Assistant recognizes that a task is important, such as logging in to a certain site or completing a purchase from your shopping cart, it will pause and ask for permission before proceeding. The Assistant is designed to operate under the same principles a human assistant would, making some minor decisions without interrupting you while knowing to ask permission for more important decisions before acting on your behalf. 

Actions are answers: Assistants unlock more questions

By prioritizing visibility, adapting to your preferences, and requesting permission before sensitive actions, the Comet Assistant works at the speed of autonomy while keeping you firmly in control. 

Transparency, user control, and sound judgment aren’t specific to Comet. As the capabilities of AI expand, and we continue to innovate toward even more useful assistants, these principles remain constant. They are the foundation of every agentic experience we’ve built, from Deep Research, to Labs, Email Assistant and Background Assistants.

We know everyone will use their AI assistants differently. The Comet Assistant is private, personal, and powerful. For you to get the most out of it, Perplexity must build it to be worthy of your trust. That’s a responsibility we take very seriously.  

As AI models advance and Comet Assistant’s features become more robust, earning and maintaining your trust transforms your relationship with your assistant from “What can it do?” to “What do I want?”  Every dream begins with a question, and your AI assistants will soon be the most powerful tool for making it come true. The future belongs to the curious, and we can’t wait to see what you’ll ask.

Comet Assistant puts you in control

Almost immediately after introducing Comet to the world, it was clear the Comet Assistant would set the standard for AI assistants, or agentic AI, on the web.  The Comet Assistant was an immediately popular tool that surprised even us. Today we’re announcing some upgrades to Comet Assistant.

As the world increasingly looks to Perplexity for leadership in agentic AI, it might be useful to outline our approach to building AI assistants.  

It’s the same approach for every assistant we build, whether it is the email assistant, background assistants, the new Comet Assistant, or another we’re excited to launch in the coming days. Your AI assistant is your AI assistant. It’s not the AI assistant, nor Perplexity’s AI assistant. It belongs to you, and we built it to power your curiosity and assist with anything you ask.

For this to work, our approach to AI assistants is grounded in three core principles: transparency in every action, user control over the assistant’s approach to assignments, and sound judgment by the assistant, knowing when to solicit feedback before taking action. All of this must take place on the bedrock of accurate AI that Perplexity builds to power answers for both human and agentic decision-making.  

Every improvement to the Comet Assistant advances these principles. It now shows you exactly what it’s doing, asks how you want it to act, and demonstrates sound judgment before completing sensitive actions. 

Transparency: Watch Comet Assistant work 

Trustworthy assistants show their work. Now it’s easier to see when the Comet Assistant is browsing for you and what actions it’s taking on your behalf. You can watch while Comet works or see its reasoning step-by-step in the Assistant sidecar. With the newest Comet Assistant, you’ll see exactly where it’s clicking, scrolling, or interacting with a site. And there are clear buttons to stop the Assistant or provide more guidance.

User Control: Decide when Comet Assistant acts 

AI assistants, like human assistants, must be trustworthy. Now, when you ask a question in the omnibar or Perplexity Search, Comet may ask if you want your Assistant to handle the task itself. You can choose to browse yourself, give Comet permission to browse on your behalf once, or let the Assistant automatically browse for you when it detects the chance to be useful. 

Sound judgment: Built-in safeguards for high-stakes decisions 

A great assistant knows some decisions require your input. When Comet Assistant recognizes that a task is important, such as logging in to a certain site or completing a purchase from your shopping cart, it will pause and ask for permission before proceeding. The Assistant is designed to operate under the same principles a human assistant would, making some minor decisions without interrupting you while knowing to ask permission for more important decisions before acting on your behalf. 

Actions are answers: Assistants unlock more questions

By prioritizing visibility, adapting to your preferences, and requesting permission before sensitive actions, the Comet Assistant works at the speed of autonomy while keeping you firmly in control. 

Transparency, user control, and sound judgment aren’t specific to Comet. As the capabilities of AI expand, and we continue to innovate toward even more useful assistants, these principles remain constant. They are the foundation of every agentic experience we’ve built, from Deep Research, to Labs, Email Assistant and Background Assistants.

We know everyone will use their AI assistants differently. The Comet Assistant is private, personal, and powerful. For you to get the most out of it, Perplexity must build it to be worthy of your trust. That’s a responsibility we take very seriously.  

As AI models advance and Comet Assistant’s features become more robust, earning and maintaining your trust transforms your relationship with your assistant from “What can it do?” to “What do I want?”  Every dream begins with a question, and your AI assistants will soon be the most powerful tool for making it come true. The future belongs to the curious, and we can’t wait to see what you’ll ask.

Comet Assistant puts you in control

Almost immediately after introducing Comet to the world, it was clear the Comet Assistant would set the standard for AI assistants, or agentic AI, on the web.  The Comet Assistant was an immediately popular tool that surprised even us. Today we’re announcing some upgrades to Comet Assistant.

As the world increasingly looks to Perplexity for leadership in agentic AI, it might be useful to outline our approach to building AI assistants.  

It’s the same approach for every assistant we build, whether it is the email assistant, background assistants, the new Comet Assistant, or another we’re excited to launch in the coming days. Your AI assistant is your AI assistant. It’s not the AI assistant, nor Perplexity’s AI assistant. It belongs to you, and we built it to power your curiosity and assist with anything you ask.

For this to work, our approach to AI assistants is grounded in three core principles: transparency in every action, user control over the assistant’s approach to assignments, and sound judgment by the assistant, knowing when to solicit feedback before taking action. All of this must take place on the bedrock of accurate AI that Perplexity builds to power answers for both human and agentic decision-making.  

Every improvement to the Comet Assistant advances these principles. It now shows you exactly what it’s doing, asks how you want it to act, and demonstrates sound judgment before completing sensitive actions. 

Transparency: Watch Comet Assistant work 

Trustworthy assistants show their work. Now it’s easier to see when the Comet Assistant is browsing for you and what actions it’s taking on your behalf. You can watch while Comet works or see its reasoning step-by-step in the Assistant sidecar. With the newest Comet Assistant, you’ll see exactly where it’s clicking, scrolling, or interacting with a site. And there are clear buttons to stop the Assistant or provide more guidance.

User Control: Decide when Comet Assistant acts 

AI assistants, like human assistants, must be trustworthy. Now, when you ask a question in the omnibar or Perplexity Search, Comet may ask if you want your Assistant to handle the task itself. You can choose to browse yourself, give Comet permission to browse on your behalf once, or let the Assistant automatically browse for you when it detects the chance to be useful. 

Sound judgment: Built-in safeguards for high-stakes decisions 

A great assistant knows some decisions require your input. When Comet Assistant recognizes that a task is important, such as logging in to a certain site or completing a purchase from your shopping cart, it will pause and ask for permission before proceeding. The Assistant is designed to operate under the same principles a human assistant would, making some minor decisions without interrupting you while knowing to ask permission for more important decisions before acting on your behalf. 

Actions are answers: Assistants unlock more questions

By prioritizing visibility, adapting to your preferences, and requesting permission before sensitive actions, the Comet Assistant works at the speed of autonomy while keeping you firmly in control. 

Transparency, user control, and sound judgment aren’t specific to Comet. As the capabilities of AI expand, and we continue to innovate toward even more useful assistants, these principles remain constant. They are the foundation of every agentic experience we’ve built, from Deep Research, to Labs, Email Assistant and Background Assistants.

We know everyone will use their AI assistants differently. The Comet Assistant is private, personal, and powerful. For you to get the most out of it, Perplexity must build it to be worthy of your trust. That’s a responsibility we take very seriously.  

As AI models advance and Comet Assistant’s features become more robust, earning and maintaining your trust transforms your relationship with your assistant from “What can it do?” to “What do I want?”  Every dream begins with a question, and your AI assistants will soon be the most powerful tool for making it come true. The future belongs to the curious, and we can’t wait to see what you’ll ask.

Share this article