What is Comet? A browser for reading and comparing web sources
Understand Perplexity Comet, its Assistant, availability and data handling before comparing public sources.
Last updated
Menus and supported features can change. Compare the source-check date below with your current app version.

Opening three official guides often leads to the same small problem: which tab explained the platform requirements, and where was the statement about data handling? Perplexity Comet is a browser that lets you ask an assistant about web content while you read. For a first exercise, BumbleV would start with a few public pages and a checked comparison table, rather than handing over email or purchases.
This introduction is based on official material checked on October 28, 2026. We have not installed Comet or generated an answer in it, and the exact desktop build has not been verified. The scope is the current online documentation, not a tested release. If your interface differs, check the installed version and current help together.
A browser with an assistant, not the name of a model
The getting-started documentation describes Comet as a Chromium-based browser with Perplexity-powered AI features. Think of the browser as the place where sources are opened and the Assistant as the part that responds to a request using relevant context. This establishes an AI-enabled application; it does not establish that the application itself was developed using AI. Official getting-started guide
Perplexity Computer and Personal Computer are separate product names. Mixing them into this tutorial would blur the installation and permission requirements. This series concerns comparing public tabs in Comet. It does not teach desktop-wide control or searching local files, and the exercise does not require connecting unrelated accounts.

Who might find this task useful?
People reading product documentation, public policies or instructions across several pages may want help organizing a first comparison. One manageable task is to identify where three documents discuss availability and data use, and explicitly mark information they do not contain. The useful outcome is not merely a smooth summary. It is a record whose claims can be traced back to the source.
A citation is a starting point for checking, not the end of checking. You still need to open the page and locate the passage supporting the statement. If a short page contains the one number you need, reading it directly may involve fewer steps. Writing a request, waiting for an answer and verifying it also take effort. We have not measured a general time saving from this workflow.
Separate availability from the cost of AI features
The current official homepage lists Mac, Windows, iOS and Android. The companion tutorial concerns desktop use; it does not assume that mobile controls are identical. Choose the official download for your operating system and inspect the actual installer requirements. Linux installation is outside the verified scope. Official Comet page
An official changelog dated October 2, 2025 announced free downloads for everyone. That is historical availability evidence, not an October 2026 launch. Free installation does not establish unlimited access to every AI capability. Check the allowances and upgrade terms actually shown for your account. We have not inspected a signed-in plan, remaining credits or a bill. Historical availability announcement
This is an owner-selected learning topic rather than a claim that Comet is a newly launched product or the most searched browser in Korea. Korean search volume and relative search-interest values have not been measured. A historical company statement about a waitlist cannot be substituted for current active users or demand.
Why begin with public pages?
Before asking about a page, understand what context may be involved. The current privacy documentation explains that requests can send relevant tab context to Perplexity. Something being open in your browser does not imply that all processing stays on your device. Public pages let you practice the comparison without introducing a private inbox or internal document into the exercise. Assistant privacy and data use
The practical question is which material you are allowed to use, where it may go and whether the resulting record is appropriate to share. This is not a reason to avoid all documents. It is a reason to choose a clear, limited input before building a workflow around it. The companion exercise uses only three public official pages.
Make the first deliverable a small verification record
A useful first table can contain the document title, its scope, missing information, the source address and a manual check status. It should help you reopen the source later, not merely reproduce a persuasive answer. If another reader can follow the addresses and understand unresolved cells, the record has done something useful.
Do not expand the task into an automatically refreshed price database before the first table is checked. Changing information needs a new observation record, while old observations remain dated. Keep the generated draft and your corrections separate enough to see what you changed. If an answer introduces a detail absent from the selected documents, mark it unresolved instead of quietly treating it as sourced.
Record the final page address as well as its title. An old help link can redirect to a replacement page, even when a search result still shows the earlier text. Keeping both gives a later check something specific to compare.
The three-tab tutorial covers preparation, the documented access points, a complete bounded request and manual preservation of the result. The documentation-based workflow comparison considers the same input when read manually. Neither companion is a hands-on benchmark or a performance verdict.