Gen AI

AI Browsers: Working Smarter

Generative AI Essentials · Hands-on micro-session

Session Objective

AI browsers like Perplexity Comet and Atlas are not just "better search bars." They collapse the gap between your browser and your AI by putting a sidecar assistant directly inside the browser window. The AI can see what you are looking at, ask questions about it, and in agent mode, click and navigate on your behalf. The key concept is the copy-paste tax - the time you spend jumping between browser tabs and AI tools. AI browsers eliminate most of that friction. We compared the old way (manual tab-switching) to the new way (a single prompt in Comet) using a real flight research task, and found that Comet's live search database produced more accurate results than Atlas, which was faster but used estimated data.

Your Challenge

Complete the following steps

STEP 1

Choose your task (1 min)

Pick one task you do regularly that involves searching for information across multiple sources. Examples: competitor pricing, industry news, supplier options, event venues, travel planning, regulatory updates, job…

STEP 2

Run it in an AI browser (7 min)

Open Perplexity Comet (comet.perplexity.ai) or Atlas (atlas.ai). Write a single prompt that covers your full research request. Let the AI browser run. Observe what it does - which sites it visits, how it combines…

STEP 3

Document the comparison (2 min)

Note how long the AI browser task took versus how long the old way would have taken. Note one thing it got right that surprised you. Note one thing it got wrong or that you had to verify.

What to Submit to Your AI Buddy

The prompt you used (full text)

The task it was for and why you do it regularly

Time estimate - old way vs AI browser way

One result that was accurate and useful

One result you had to verify or that was wrong

One sentence on how this changes your research workflow

Grading Criteria

The prompt is task-specific,

A strong prompt tells the AI browser what you are trying to accomplish, not just what to search for. "Find me a hotel in KL" is generic. "I have a 2-night business trip to KL in April, budget under $150 per night, I need a hotel near the KL

The task is a real recurring

The challenge asks for something you actually do regularly. "Find the population of France" is not a recurring work task. "Track competitor pricing for our top three product categories every Monday" is. Real tasks produce real insights abou

The comparison is honest

A strong submission acknowledges both what worked and what did not. AI browsers make mistakes. Saying "it was perfect" is less useful than saying "it found the right sites but got one price wrong - I verified on the airline's website." Hone

The workflow implication is

Your submission ends with a specific observation about how this changes your work. Not "AI is cool" - something concrete. "This replaces 20 minutes of tab-switching every Monday morning" or "I still need to verify pricing but the first pass

Next → AI Shopping: The End of Browse and Click

Micro-Session Resources