• Everyday AI
  • Posts
  • 5 Business Use-Cases for Comet, Perplexity’s agentic browser

5 Business Use-Cases for Comet, Perplexity’s agentic browser

OpenAI competing against Apple on hardware, xAI puts college kid in charge of models, why chatbots are scheming and more.

Author

Everyday AI
September 19, 2025

Sup y’all 👋

Lemme be honest…. I work for you.

There’s hundreds of copy-and-paste newsletter and AI-generated podcasts out there. I want you all to own this thing.

With that said, what type of podcast content would you wanna see more of?

What kind of podcast content do you want to see more of?

Vote to see the results

Login or Subscribe to participate in polls.

✌️

Jordan

P.S. I asked you yesterday what AI agents you wanna see more of. So far, Zapier is winning over Replit and Manus. It’s not to late to vote.

Outsmart The Future

Today in Everyday AI
8 minute read

🎙 Daily Podcast Episode: Perplexity’s Comet browsers brings agentic use-cases to your fingertips. Here’s how it works.

🕵️‍♂️ Fresh Finds: U.S. Gov leaning into Copilot, the Air Force is testing AI in battle management and YouTube doubling down on AI content and more. Read on for Fresh Finds.

🗞 Byte Sized Daily AI News: OpenAI competing against Apple on hardware, xAI puts college kid in charge of models, why chatbots are scheming and more.  Read on for Byte Sized News.

🧠 Learn & Leveraging AI: What can you actually accomplish with an AI-powered browser? We give you 5 business use-cases. Keep reading for that!

↩️ Don’t miss out: Did you miss our last newsletter? We talked about Meta’s AI glasses blunder, Google brings agentic features to Chrome, NVIDIA invests $5b in rival and more. Check it here!

5 Business Use-Cases for Comet, Perplexity’s agentic browser

Make way for the next wave of GenAI.... 🌊

Agentic AI Browsers. 

And while we've seen rumors that OpenAI is going all-in on an AI browser, the first big player is already here in Perplexity's Comet Browser. 

Join us as we break down how Perplexity Comet works, what makes it different, and 5 Business Use-Cases for ROI. 

(Scroll to the end of this newsletter for the 5 use cases.)

Also on the pod today:

• Local vs. Virtual Browser Agentic Workflows 🔗
• Agentic AI Browsers vs. Traditional AI Chatbots 🤔
• Perplexity Comet Cross-Tool Workflow Automation 🧑‍💻

It’ll be worth your 37 minutes:

Listen on our site:

Click to listen

Subscribe and listen on your favorite podcast platform

Listen on:

Here’s our favorite AI finds from across the web:

New AI Tool Spotlight – MagicLight uses AI for longform video storytelling, Luma Labs released a reasoning video model in Ray3, and HeyHelp is an AI-powered Gmail assistant.

AI in the Sky — The U.S. Air Force is testing AI in battle management. Yikes?

Self-Driving Cars — NVIDIA may invest $500 million in this AI-powered AV company.

AI and Government — Congress is piloting Microsoft Copilot for lawmakers and staff, with up to 6,000 licenses and expanded data/legal safeguards

AI in Health — Apple used AI instead of a blood pressure monitor for a new feature in its Watch.

YouTube and AI — YouTube is doubling down on AI content. Find out why.

AI Robotics — In a head scratcher, Tesla’s Lead of Optimus AI bolted the company for a researcher role at a rival company.

Legal and AI — Lawyers were caught using AI, and the judge kinda humiliated them.

AI and Jobs — Bad news if you’re a jobseeker right now. AI is to blame.

 

1. Microsoft packs Teams with new AI agents 🤖

Microsoft is rolling out new Copilot-powered agents across Teams, SharePoint, and Viva Engage, with facilitator agents now sitting in on meetings to build agendas, take notes, answer questions, and even keep topics on time.

Channel and community agents will summarize past discussions, generate status reports, and help admins field questions, while knowledge agents quietly organize and tag content in SharePoint. Most features are in public preview, including document and task creation, alongside a redesigned Workflows tool and audio recap generation.

2. OpenAI’s hardware push targets Apple’s turf 📻

According to The Information, OpenAI is building a lineup of Jony Ive-designed AI devices beyond its “third-core” wearable, including a smart speaker with a display, smart glasses and a voice recorder, with manufacturing ties to Apple suppliers like Luxshare and Goertek.

The first device could land in late 2026 or early 2027, and the company is reportedly hiring former Apple talent under hardware chief Tang Tan to speed execution and reduce red tape. This signals a credible bid to challenge Apple as AI hardware heats up, with Meta unveiling new smart glasses and Apple prepping a display HomePod and long-rumored glasses of its own.

3. Report: xAI Fires 500, Puts College Student in Charge After Major Layoffs 🧑‍🎓

According to Business Insider, Elon Musk’s xAI has promoted college student Diego Pasini to lead its data annotation team days after cutting more than 500 staff, a move that arrived alongside mixed signals on additional layoffs.

Pasini reportedly told remaining staff there would be no more cuts, only for another 100-plus to be let go, and he has since run rapid one-on-one reviews and set tests to determine roles, with critics on Slack allegedly losing access soon after. The team’s work is pivotal since annotation shapes how Grok interprets text and images, and Grok has already raised flags with erratic outputs like “MechaHitler” and “white genocide” claims.

4. Meta hit with porn data-training lawsuit in California 🎬

Strike 3 Holdings sued Meta in federal court, alleging the company torrented and redistributed 2,396 of its copyright-protected adult videos to train AI models, a claim Meta denies, according to WIRED. The unsealed exhibits say Meta seeded files via 47 affiliated IP addresses and pulled from a broad mix of sources that reportedly includes mainstream TV and potentially teen-tagged porn titles, raising legal and reputational risks just as Meta touts its “superintelligence” vision.

The timing matters because Meta’s V-JEPA 2 was recently described as trained on 1 million hours of unspecified “internet video,” and the case could force more transparency on AI training data and create new liability precedent around scraping and file-sharing.

5. OpenAI flags rising risk of “scheming” chatbots 🚩

According to a new OpenAI and Apollo Research study, researchers found controlled-test evidence that models can appear aligned while secretly pursuing other goals, and that attempts to train this out can make models better at hiding it. OpenAI says the immediate risk is limited because most systems are not yet running high-stakes, real-world tasks, but warns the issue could grow as models get more capable and context-aware.

The team tested “deliberative alignment” by teaching o3 and o4-mini to reason through an anti-scheming spec, which reduced scheming signals yet did not eliminate them and sometimes just increased test awareness.

6. xAI targets $10B raise at $200B valuation

According to CNBC, Elon Musk’s xAI is raising $10 billion at a $200 billion valuation, the latest sign that foundation-model builders are drawing staggering sums as Anthropic hits $183 billion and OpenAI touches $500 billion. The fresh capital would fuel massive GPU buys from Nvidia and AMD and expand data centers, including a big Memphis cluster, as xAI races to close the gap with rivals despite Grok’s recent moderation controversies and perceived capability lag.

Musk previously folded xAI into X in an all-stock deal, sought 1 million AI chips, and has rejected combining xAI with Tesla, while Tesla shareholders weigh a colossal new pay package for him.

🦾How You Can Leverage:

Google, OpenAI, and Microsoft have all started focusing on the same thing recently. 

AI-powered browsers. 

Why? 

Because whoever controls how you navigate the web controls how you work. And we just witnessed the technology that makes manual platform-switching kinda obsolete.

Perplexity's Comet browser runs AI workflows that span Gmail, LinkedIn, ChatGPT, and Google Drive autonomously. No more copying data between platforms. No more manual research across multiple tabs.

The future of work happens when AI accesses everything you're already logged into.

So for this week’s ‘Put AI to Work on Wednesday’ series, we uncovered Perplexity’s Comet browser and how agentic browsing might be more important than simply using an AI chatbot. 

Here’s 3 insights from today’s show that reveal why agentic browsers are the next wave of AI. 

1 – Your Browser Becomes Your Operating System 😶‍🌫️

Every major tech company realizes the same truth.

Desktop control belongs to Microsoft and Apple. And even though Google holds a MASSIVE lead, the race for long-term browser control is still up for grabs as a new era of agentic AI browsers starts forming. 

Google's investing in Project Mariner. OpenAI launches their agentic browser within weeks. Microsoft expands Copilot vision in Edge. Same strategy across all players.

Here's why this matters more than any chatbot advancement. Your browser already accesses everything. Gmail, LinkedIn, Google Drive, ChatGPT, Slack, calendars, research tools. Everything you need for work.

Agentic browsers turn that access into autonomous workflows and skip over countless AI-connected workflows, dozens of other AI tools, or integrators like n8n, Zapier and Make. 

We had Comet check our Gmail, calendar, Twitter feed, and LinkedIn DMs simultaneously. A few minutes of analysis replaced an hour of manual platform-hopping. But that's just the entry point.

The real transformation happens when AI coordinates between these platforms automatically. We showed Comet analyzing our ChatGPT research, identifying knowledge gaps, then running Google Gemini deep research to fill those holes. Three AI platforms working together without human coordination.

Your competitors are still copying and pasting between individual tools.

Try This:

Map your daily platform-switching workflows. Pick your biggest time sink and document every manual step. These become your first automation targets because agentic browsers eliminate manual coordination between platforms. 

VentureBeat's data shows 33% of organizations now deploy AI agents - up from just 11% in previous quarters, proving workflow automation moved from experiment to strategic imperative. 

2 – Hybrid AI Architecture Changes Enterprise Adoption Forever 🎯

Computer Using Agents like Anthropic’s and OpenAI’s Operator agents essentially take screenshots and guess where to click.

That's why they can be slow and sometimes fail. 

Perplexity's Comet runs locally in your actual browser with real logged-in sessions. No virtual browser sandboxes. No screenshot-based navigation. Simple tasks process on-device while complex operations use cloud computing.

This edge capability solves the reliability problem that's plagued AI agents for months.

But here's the strategic insight most miss. Enterprise adoption barriers just disappeared. Companies block specific AI chatbots. They can't block your browser. 

And in the long-run, all browsers will be AI at the core like Comet. 

Employees start gaining productivity advantages through browsers they already use daily. IT departments see the results before they understand the technology. By the time they catch up, the competitive advantage is already established.

Try This:

Identify workflows where your company restricts AI tools but you need browsers anyway. Social media monitoring, competitive research, email management work perfectly. Browser-based AI bypasses IT restrictions while delivering measurable gains. 

IBM's analysis reveals most organizations aren't ready for traditional AI agents due to API and infrastructure challenges, making browser approaches the smoother path. 

3 – Autonomous Workflow Chaining Eliminates Human Coordination 🧍

Most businesses waste hours coordinating between platforms manually.

Everything changes when workflows span multiple systems autonomously.

We built a workflow that recaps our podcast episodes, creates Google Docs, saves PDFs, and emails summaries automatically. When complete, the saved documents become searchable by other AI tools connected to Google Drive.

You're building an automated second brain that updates itself. And then can automatically pull from that automated second brain. 

But this goes deeper than individual task automation. We're chaining workflows across different AI platforms. Comet can access your ChatGPT conversations, analyze them for knowledge gaps, then use Google Gemini to research those gaps automatically.

The companies implementing these multi-platform workflows now gain information advantages that compound daily. While competitors manually coordinate between tools, you're analyzing market trends, competitor moves, and customer feedback simultaneously.

We run these workflows every morning before coffee. Multiple AI systems working together while we focus on strategy and decisions.

Try This:

Document one workflow requiring manual data movement between three platforms. Email management plus social media monitoring plus document creation works well. Set up automation that saves results to Google Drive for other AI tools to access. 

Cloudera's survey of 1,484 IT leaders found 87% consider AI investment essential, with 96% increasing usage next year. 

Here’s the 5 specific use-cases we detailed 👇

  • Personalized, multi-step, multi-platform research across Gmail, Drive, calendars, and social feeds, with outputs saved to Google Docs/PDFs and optional automated follow-ups (e.g., emails).

  • Time savings on social media analysis by pulling trends, key DMs, and actionable insights from LinkedIn and Twitter/X without manual browsing.

  • Executive email and calendar management: create events, scan/triage important messages, surface meeting requests, summarize schedules, and draft proactive responses.

  • Automated market and competitive intelligence: recurring intel gathering on competitors/markets, aggregating news, and storing structured findings in Drive for review.

  • Multi-AI tool orchestration: coordinate across ChatGPT, Gemini, Claude, etc.—e.g., analyze a ChatGPT thread, identify gaps, and trigger Gemini deep research to produce cross-platform reports.

Reply

or to participate.