Stephen introduces the video by sharing his initial skepticism about AI browsers that transformed into enthusiasm after using Perplexity Comet for a week. He positions Comet as superior to alternatives like Dia, Google Chrome with Gemini, and ChatGPT Agent Mode, particularly in its ability to take action on websites.

He previews the 10 real-world use cases he'll demonstrate in the video, including finding promo codes, YouTube features, creating product lists, automating Apple Podcasts tasks, finding files on websites, analyzing metrics, and more. He briefly acknowledges privacy concerns but promises to address them at the end of the video.

The first use case demonstrates how Perplexity Comet can find and test promo codes for online shopping. Stephen conducts a side-by-side comparison with ChatGPT in agent mode, showing how Comet can not only find potential discount codes but actually test them in real-time within the cart on websites.

The key advantage highlighted is speed and automation - where Comet found and applied a working promo code in seconds, ChatGPT took 11 minutes to accomplish the same task. Stephen also shows how Comet can multitask, running one promo code search while starting another in a new tab.

The second use case showcases Comet's powerful YouTube integration capabilities. Stephen demonstrates how the AI browser can find specific moments in videos, like Neil Armstrong's famous quote or Steve Jobs discussing Apple at the intersection of liberal arts and technology, and immediately play those exact segments.

Beyond finding specific clips, Stephen shows how Comet can analyze video content in real-time while watching - identifying what equipment a creator is using, finding timestamps for specific topics, providing actionable tips from interview content, extracting thumbnail images, and even creating playlists by searching for and adding relevant videos automatically.

In this section, Stephen demonstrates how Perplexity Comet can help create clean, formatted product lists from multiple Amazon tabs. He shows how the browser can take all open Amazon tabs, extract the essential product information, create shortened URLs, and format everything into a clean list.

Stephen also mentions how Text Expander has become an invaluable complementary tool for his AI browser workflow, allowing him to store and quickly access common prompts for both Comet and ChatGPT through simple keyboard shortcuts.

Stephen shares a niche but highly valuable use case for his podcast workflow. When uploading subscriber-only episodes to Apple Podcasts, the platform requires users to wait for processing to complete before manually clicking the publish button - a tedious monitoring task.

Perplexity Comet solves this problem by automating the monitoring process. Stephen simply instructs Comet to refresh the page every minute and click the publish button once it becomes available. This was actually his first discovered use case for the browser and remains one he uses daily for his podcast, Primary Technology.

This section demonstrates Comet's ability to quickly locate specific files or information on websites with complex navigation. Stephen shows how the browser can find and download forms like a W9 from the IRS website without having to navigate through multiple pages.

The capability extends to navigating complex government websites like DMV portals. Rather than manually searching through confusing navigation menus, Stephen can simply ask Comet what documentation he needs for vehicle registration, and the browser will not only provide the information but also navigate directly to the relevant page.

Stephen demonstrates how Perplexity Comet can analyze and interpret data from analytics platforms. Using his podcast hosting platform Transistor as an example, he shows how Comet can answer specific questions about download metrics, such as identifying the second-highest downloaded episode or which month had the most downloads.

The feature extends to more complex analytics platforms like Google Search Console and Google Analytics. Stephen explains that Comet can navigate these interfaces, adjust date ranges, download CSV files if needed, and extract precisely the information requested - eliminating the need to manually filter and analyze the data.

Building on the previous analytics capabilities, Stephen specifically highlights how Comet excels at analyzing YouTube Studio metrics. He demonstrates how the browser can change date ranges in YouTube's advanced analytics mode and extract specific performance data like monthly watch hours.

A particularly useful feature is Comet's ability to format extracted data into clean tables. Stephen shows how the browser can present YouTube metrics in a well-structured markdown table that can be easily copied and used elsewhere.

Stephen demonstrates how Perplexity Comet can assist with spreadsheet tasks by analyzing data and providing formulas. Using the YouTube metrics table from the previous section, he shows how Comet can calculate statistics like average watch time directly from the spreadsheet data.

While noting that Comet's direct interaction with spreadsheets is currently limited, Stephen highlights its ability to generate useful spreadsheet formulas. He shows how the browser can create formulas for tasks like calculating column sums, which users can then copy and paste into their spreadsheets.

This section showcases Comet's superior ability to search for specific content within websites. Stephen demonstrates this using his podcast website, showing how Comet can find episodes where specific topics like the Mac Studio are discussed without requiring traditional site search or Google site-specific searches.

The standout capability is Comet's ability to not just find content but navigate directly to the relevant sections. Stephen shows how after finding an episode mentioning Mac Studio, he can ask Comet to navigate to the specific part of the transcript where the topic is discussed, saving significant time compared to manual searching.

Stephen demonstrates Comet's powerful LinkedIn search capabilities, referencing Wall Street Journal tech columnist Joanna Stern's endorsement of this feature. He shows how Comet can find specific people matching complex criteria, such as someone who used to work on Apple's AI team but now works at Meta.

The demonstration highlights how Comet can execute sophisticated multi-parameter searches and navigate directly to the profiles of the people who match the criteria, saving time compared to using LinkedIn's native search filters.

Stephen provides two bonus use cases beyond his primary ten examples. First, he demonstrates how Comet can search and filter real estate listings on Zillow based on natural language criteria (homes in Lakeland, Florida under $400,000 with specific bedroom/bathroom requirements), eliminating the need to manually set multiple filters.

As a lighthearted second example, Stephen tests whether Comet can play Wordle, showing a simultaneous race between himself and the AI browser to solve the daily word puzzle. While acknowledging that playing Wordle is something he'd prefer to do himself, it demonstrates Comet's ability to interact with games and puzzles.

In the final section, Stephen addresses the privacy concerns associated with Perplexity Comet. He acknowledges that Perplexity's CEO has stated they track user activity in the browser to sell personalized ads, which Stephen admits "doesn't feel great."

Despite these concerns, Stephen explains his personal risk calculation. He won't use Comet for sensitive activities like online banking but finds the productivity benefits worth it for specific tasks like automating podcast publishing or finding YouTube videos. He expresses hope that other browsers like Dia or Apple's Safari might eventually incorporate similar capabilities with better privacy protections.

Perplexity Comet represents a significant evolution in how AI can enhance everyday browsing and productivity tasks. While traditional browsers require users to perform actions manually and sequentially, Comet's ability to understand context, navigate websites, and perform complex actions autonomously allows for a fundamental shift in workflow - enabling background task automation while users focus on other activities.

However, this advancement comes with important privacy trade-offs that users must consider. The stated intention to track browsing activity for ad targeting creates a clear tension between convenience and privacy. Users need to make informed decisions about which tasks are appropriate for AI browsers versus traditional browsers with stronger privacy protections.

So what? As AI browsers continue to evolve, we're witnessing the early stages of a transformation in how we interact with the web. Rather than manually navigating, searching, and performing repetitive tasks, we can increasingly delegate these actions to AI assistants integrated directly into our browsing experience. The key for users is finding the right balance - leveraging these powerful capabilities for appropriate tasks while maintaining privacy for sensitive activities.