Close Menu
    Main Menu
    • Home
    • News
    • Tech
    • Robotics
    • ML & Research
    • AI
    • Digital Transformation
    • AI Ethics & Regulation
    • Thought Leadership in AI

    Subscribe to Updates

    Get the latest creative news from FooBar about art, design and business.

    What's Hot

    LLM Embeddings vs TF-IDF vs Bag-of-Phrases: Which Works Higher in Scikit-learn?

    March 3, 2026

    Understanding Audio Annotation for Speech Recognition Fashions

    March 3, 2026

    Uncensy Picture Generator Costs, Capabilities, and Characteristic Breakdown

    March 3, 2026
    Facebook X (Twitter) Instagram
    UK Tech InsiderUK Tech Insider
    Facebook X (Twitter) Instagram
    UK Tech InsiderUK Tech Insider
    Home»Machine Learning & Research»Are AI Browsers Any Good? A Day with Perplexity’s Comet and OpenAI’s Atlas
    Machine Learning & Research

    Are AI Browsers Any Good? A Day with Perplexity’s Comet and OpenAI’s Atlas

    Oliver ChambersBy Oliver ChambersNovember 27, 2025No Comments11 Mins Read
    Facebook Twitter Pinterest Telegram LinkedIn Tumblr Email Reddit
    Are AI Browsers Any Good? A Day with Perplexity’s Comet and OpenAI’s Atlas
    Share
    Facebook Twitter LinkedIn Pinterest Email Copy Link


    Are AI Browsers Any Good? A Day with Perplexity’s Comet and OpenAI’s AtlasPicture by Writer

     

    # Introduction

     
    Many know-how specialists will let you know that AI browsers are the way forward for work. The truth is much extra difficult. After spending weeks testing Perplexity’s Comet alongside ChatGPT’s Atlas browser, I found one thing uncomfortable. These instruments can prevent time on particular analysis duties, however they crumble on something advanced. For information scientists and analysts already overwhelmed by a number of open tabs, that is essential. However earlier than you select Chrome over an AI browser, you should perceive what these instruments truly do and, extra importantly, what they can not do.

     

    # Understanding AI Browsers

     
    An AI browser feels like one of the best browser till you truly use one. At its core, it’s a net browser, much like Chrome or Safari, however with a man-made intelligence (AI) assistant built-in right into a sidebar. This assistant can learn any webpage you’re viewing, reply questions on it, summarize articles, and even try to automate duties — e.g. filling out kinds or evaluating costs throughout web sites.

    Perplexity’s Comet is completely different from a conventional browser as a result of it understands context throughout a number of tabs directly. If in case you have analysis papers, information dashboards, and competitor stories open concurrently, Comet’s assistant can learn all of them collectively and synthesize what it finds. The browser was constructed on Chromium, the identical basis as Google Chrome, so it helps the identical extensions you already use.

     

    Perplexity Comet Landing pagePerplexity Comet Landing page
    Picture by Writer

     

    The important thing promoting level is automation. The Comet Assistant sits in a sidecar panel and might carry out multi-tab operations. As a substitute of manually copying information from 5 web sites right into a spreadsheet, you simply ask Comet to do it for you.

     

    # Making use of Comet in a Information Scientist’s Workflow

     
    To grasp if AI browsers matter to your workflow, let me stroll by way of what truly occurred when an information science skilled spent a day with Comet.

     

    // Using Comet for Market Analysis

    If a researcher wanted aggressive information on machine studying platforms, usually, this includes opening 10 to fifteen tabs and manually extracting pricing, options, and efficiency metrics right into a spreadsheet.

    With Comet, the researcher created a workspace (Comet calls it a Area) and opened tabs for six competing platforms. Then they requested the assistant:

    “Create a comparability desk with platform title, pricing per 30 days, free tier options, and newest replace date.”

     

    The assistant will learn all six web sites concurrently and ship a correctly formatted desk in three minutes. This process takes roughly 45 minutes manually.

     

    // Using Comet for Analysis Synthesis

    Subsequent got here analyzing three tutorial papers on anomaly detection algorithms. Let’s say the duty is to extract key mathematical ideas, evaluate methodologies, and determine which method may work finest for his or her dataset.

    With conventional searching, this implies opening every PDF, studying, taking notes, switching between tabs, and manually evaluating. They requested Comet:

    “Summarize the core mathematical distinction between these three papers and rank them by how nicely they might work with real-time monetary information.”

     

    The assistant synthesized throughout all three open papers and delivered a ranked abstract with direct quotes from every supply. The assistant even understood the context from their earlier questions, so it prioritized monetary information functions with out being requested twice.

    Guide time usually required 2 hours. Precise time: 18 minutes.

     

    // Figuring out Comet’s Limitations in Analysis Synthesis

    Right here is the place Comet breaks down. They wanted the assistant to mechanically extract information from a JavaScript-heavy dashboard, validate it towards a SQL database, and flag mismatches. That is the place AI browsers seemingly cease being helpful.

    Comet couldn’t work together with the dynamic dashboard correctly. It tried to click on buttons, however the web page didn’t reply as anticipated. The assistant obtained confused by the interactive components and gave up. Our researcher needed to do it manually.

    That is the sincere fact. AI browsers excel at studying and synthesizing static data. They battle with advanced, multi-step interactions in fashionable net functions.

     

    # Comet Versus ChatGPT Atlas: Which Browser to Select?

     
    If in case you have been following AI information, you realize OpenAI launched ChatGPT Atlas, their very own AI browser. It performs sooner on some duties however has completely different limitations.

    ChatGPT Atlas is extra aggressive with automation. In pace exams, it accomplished a analysis temporary in 47 seconds in comparison with Comet’s 58 seconds. For CSV extraction, Atlas was considerably sooner, at 36 seconds, in comparison with Comet’s 52 seconds.
    Nevertheless, Atlas has a crucial weak spot: it at present solely works on macOS. In case your staff makes use of Home windows or Linux, you can not use it but.

    Extra importantly, Atlas is locked into the OpenAI ecosystem. If you wish to use it with different instruments or workflows, you’re out of luck. Comet integrates with Perplexity Professional, permitting it to work alongside your current AI analysis workflow.

    For information scientists particularly, the selection will depend on your major process. If you’re synthesizing data from a number of sources, Comet is healthier as a result of it maintains stronger context consciousness throughout tabs. If you’re automating repetitive form-filling or net scraping duties, Atlas may prevent barely extra time, however provided that you utilize a Mac.

    The efficiency distinction isn’t huge sufficient to make both browser important. Each have the identical elementary limitation: they battle with advanced, interactive web sites.

     

    Metric Perplexity Comet ChatGPT Atlas Courageous Leo
    Analysis Transient (sec) 58 47 62
    CSV Extraction (sec) 52 36 48
    Multi-Step Evaluate (sec) 185 132 161
    General Usefulness Excessive Very Excessive Excessive

     

    # Explaining Why AI Browsers Exhibit Uneven Efficiency

     
    Understanding the underlying know-how helps clarify why AI browsers exhibit such uneven efficiency.

    Once you ask an AI browser to summarize a webpage, it receives the complete textual content content material and processes it utilizing a language mannequin. That is basically what ChatGPT does, however with a single webpage as a substitute of a common immediate. For this process, AI is genuinely sturdy.

    However when a web site makes use of JavaScript to dynamically load content material, every thing breaks down. The browser can view the rendered web page, however the AI can’t essentially decide which components are interactive, the place the info resides, or the way to correctly set off dynamic content material. When Comet tried to make use of the JavaScript dashboard in our afternoon take a look at, it was basically attempting to resolve a puzzle it was not designed for.

    This isn’t a flaw particular to Comet or Atlas. This can be a elementary limitation of how fashionable net functions work. Till AI techniques can reliably interpret advanced, interactive interfaces in real-time, AI browsers will stay extra helpful for analysis and studying than for automation.

     

    # Addressing the Privateness and Safety Drawback of AI Browsers

     
    That is the place AI browser corporations get defensive. To be helpful, AI browser assistants require permission to learn your emails, view your calendar, entry your contacts, and work together along with your accounts. Perplexity and OpenAI want broad entry to really ship on the automation guarantees.

    That entry creates assault alternatives. Cybersecurity researchers have recognized a sort of assault often called immediate injection assaults as a systemic threat throughout all AI browsers. Right here is the way it works: a malicious web site hides directions in its web page code that trick the AI into taking actions it mustn’t take. An attacker may doubtlessly manipulate Comet into exposing your emails or making unauthorized purchases.

    There may be at present no full protection towards immediate injection. It’s an rising drawback that your complete trade is grappling with.

    For information scientists working with delicate analysis information, that is value contemplating earlier than granting your browser such intensive entry. Many organizations with strict compliance necessities (authorized corporations, monetary establishments, healthcare corporations) is not going to enable these instruments on company networks due to this threat.

     

    # Actual Productiveness Positive aspects and Actual Limitations

     
    Here’s what sincere testing truly exhibits.
    Duties the place AI browsers ship measurable time financial savings embrace:

    • Summarizing lengthy articles or stories (saves 15 to half-hour per doc)
    • Evaluating data throughout a number of static web sites (saves 30 to 60 minutes)
    • Extracting key data from PDFs (saves 20 to 45 minutes)
    • Creating analysis tables from a number of sources (saves 30 to 60 minutes)

    Duties the place AI browsers underperform or fail:

    • Working with JavaScript-heavy dashboards or interactive APIs
    • Performing multi-step, advanced interactions throughout completely different websites that require dynamic decision-making
    • Dealing with duties involving delicate firm APIs or inner networks

    One researcher on Reddit who examined Comet intensively reported that it truly doubled their productiveness for analysis synthesis, saving them roughly one hour per day. However that is for somebody doing information-intensive work. For common net searching or working with advanced functions, the time financial savings disappear.

    The Perplexity CEO claimed that Comet may enhance productiveness by 20 %, doubtlessly including trillions to GDP. That is enterprise capital discuss, not engineering actuality. Actual-world utilization exhibits extra modest features — possibly 5 to 10 % time financial savings for particular workflows, zero % financial savings for others.

     

    # Figuring out if You Ought to Swap to an AI Browser

     
    The sincere reply is, it will depend on your precise workflow.
    In case your work primarily includes studying, analyzing, and synthesizing data from a number of on-line sources, an AI browser will prevent important time. Information scientists doing literature critiques, aggressive evaluation, or analysis synthesis fall into this class.

    In case your work includes interacting with net functions, filling out kinds, or working with JavaScript-heavy instruments, it is best to stay along with your present browser. The AI browser is not going to help you and will even gradual you down.
    The associated fee issues too. Comet requires a Perplexity Professional Max subscription, which prices roughly $20 per 30 days. If you’re saving thirty minutes per day on analysis work, that roughly pays for itself. If you’re saving 5 minutes per week, it doesn’t.

    ChatGPT Atlas is at present free throughout testing, however it is going to presumably value one thing ultimately. For macOS customers conducting research-intensive work, it might grow to be worthwhile. For everybody else, the equation is much less clear.

     

    # Contemplating Privateness, Safety & the Uncomfortable Reality

     
    Earlier than adopting any AI browser, you should perceive what you’re buying and selling.
    AI browsers require sending web page content material to cloud servers for processing. This implies the info you’re viewing passes by way of exterior techniques earlier than being summarized or analyzed. For many analysis work, that is high-quality. For delicate or proprietary data, it’s a drawback.

    Some organizations deal with this through the use of AI browsers just for public analysis and conserving proprietary work in conventional browsers. That is affordable, but it surely additionally defeats a part of the purpose of getting an AI browser.

    The immediate injection assault threat is actual, however not instantly catastrophic if you happen to use frequent sense. Don’t use AI browser brokers to work together with high-stakes web sites, e.g., banking portals or delicate firm techniques. Use them for analysis, evaluation, and knowledge synthesis the place the draw back of an error is comparatively low.

     

    # Conclusion

     
    In conclusion, AI browsers aren’t going away, however they won’t substitute conventional browsers anytime quickly. The know-how will enhance. Efficiency points with JavaScript functions are doubtless to enhance. New safety requirements will ultimately handle immediate injection assaults.

    What issues now’s matching the software to the duty. For information scientists, one of the best real-world use case is precisely what we examined: synthesizing analysis data from a number of sources. That is the place AI browsers truly ship measurable productiveness features with out compromising on safety or coping with extreme complexity.

    Should you spend hours every week studying, evaluating, and synthesizing data, take a look at Comet or Atlas for every week. The time financial savings alone may justify the price. In case your work is primarily application-focused or includes advanced interactive duties, save your cash and persist with what you realize works.
     
     

    Shittu Olumide is a software program engineer and technical author captivated with leveraging cutting-edge applied sciences to craft compelling narratives, with a eager eye for element and a knack for simplifying advanced ideas. It’s also possible to discover Shittu on Twitter.



    Share. Facebook Twitter Pinterest LinkedIn Tumblr Email
    Oliver Chambers
    • Website

    Related Posts

    Construct Semantic Search with LLM Embeddings

    March 3, 2026

    Reduce Doc AI Prices 90%

    March 3, 2026

    Why Capability Planning Is Again – O’Reilly

    March 2, 2026
    Top Posts

    Evaluating the Finest AI Video Mills for Social Media

    April 18, 2025

    Utilizing AI To Repair The Innovation Drawback: The Three Step Resolution

    April 18, 2025

    Midjourney V7: Quicker, smarter, extra reasonable

    April 18, 2025

    Meta resumes AI coaching utilizing EU person knowledge

    April 18, 2025
    Don't Miss

    LLM Embeddings vs TF-IDF vs Bag-of-Phrases: Which Works Higher in Scikit-learn?

    By Yasmin BhattiMarch 3, 2026

    On this article, you’ll learn the way Bag-of-Phrases, TF-IDF, and LLM-generated embeddings examine when used…

    Understanding Audio Annotation for Speech Recognition Fashions

    March 3, 2026

    Uncensy Picture Generator Costs, Capabilities, and Characteristic Breakdown

    March 3, 2026

    The High 10 LLM Analysis Instruments

    March 3, 2026
    Stay In Touch
    • Facebook
    • Twitter
    • Pinterest
    • Instagram
    • YouTube
    • Vimeo

    Subscribe to Updates

    Get the latest creative news from SmartMag about art & design.

    UK Tech Insider
    Facebook X (Twitter) Instagram
    • About Us
    • Contact Us
    • Privacy Policy
    • Terms Of Service
    • Our Authors
    © 2026 UK Tech Insider. All rights reserved by UK Tech Insider.

    Type above and press Enter to search. Press Esc to cancel.