I’ve been playing with ChatGPT Agent and Perplexity Comet—and I’m not sure if I’m witnessing the future of productivity or just an elaborate way to make simple tasks more complicated.
The premise sounds great: tell an AI what you want, and it’ll just… do it. Book your train tickets, find that obscure book, curate a playlist based on today’s chaos. But after spending way too much time watching these things fumble around the internet, I’m starting to wonder what the problem is (it’s me, hi, I’m the problem, it’s me).
1. The Great Train Booking Experiment
I asked both agents (in German) to book me the cheapest train from Hamburg to Copenhagen for a September trip. Simple enough, right? I mean, I could knock this out in under a minute on the Deutsche Bahn website.
ChatGPT Agent found the better deal at €63.26, but watching it navigate the Bahn website was like watching someone perform surgery with oven mitts. Ten excruciating minutes of clicking cookie banners, scrolling through options, and eventually asking me to log in through its browser. The whole time I’m sitting there thinking: I could have done this already, got me coffee, and probably written half this post.
Perplexity Comet was faster—3:40 minutes—but missed the savings by booking trains separately, resulting in the same connection, but for €70.94.
2. The Book Hunt That Wasn’t
Next test: finding a copy of “We Are Anonymous” (shameless plug—it’s a book I co-authored). I switched to English for this one. ChatGPT Agent actually found a German second-hand seller and put it in a shopping cart for €5.29. Not bad, except I remembered I still have copies lying around my apartment.
Perplexity Comet gave me the “it’s out of print” speech and suggested I hunt for a “legitimate copy”—which, fair enough, but one of its links led to a completely different ebook by one of my co-authors. To be fair, the second link worked, directing me to a second-hand seller: $4.41 with free shipping in Germany.
3. News Roundup: Lost in Translation
Here’s where things get weird. I asked in English for German news in Axios-style bullet points, and both agents… used only English sources? In 2025? When AI can translate faster than I can say “Guten Tag”?
Perplexity Comet took 20 seconds and served up random news, among them week-old news about a train derailment, based on one source, Euronews:
“A deadly train derailment in southern Germany, caused by a landslide after heavy rain, has killed at least three people and injured several others. The incident has led to major disruptions on German rail lines, and investigations into safety and infrastructure are ongoing.”
ChatGPT Agent took nine minutes (nine!) but delivered actually witty, recent summaries. It appeared to read many different sources, but used only two in the end. For example this, from Reuters:
“Trade war deja vu. The EU‑US ‘deal’ agreed last month still slaps 50% tariffs on steel and aluminium while imposing 15% on most goods. German finance minister Lars Klingbeil is in Washington pushing for a quota system to spare German steel jobs. The uncertainty is already biting: Germany’s engineering orders fell 5% in June as firms plan around the new tariff regime. Why it matters: Europe’s biggest exporter can’t afford a 50% tax on metal, and without a quota the trade peace may look a lot like a war. Be smart: This isn’t 1930, but it’s a reminder that ‘tariff relief’ can still punch industries in the gut.”
4. The Spotify Playlist Fail
The next test was my attempt to somehow get through Monday: make me a Spotify playlist based on today’s news.
Perplexity Comet suggested seven tracks, including Pet Shop Boy’s “Opportunities (Let’s Make Lots of Money)” to accompany news on Tesla, tariffs, global plastic crisis. I told Comet to make the actual playlist it, but it failed to open Spotify.
ChatGPT Agent made it to Spotify’s login page and had me enter my credentials before hitting a “Playback protected content is not supported” wall. It did suggest “Masters of War” by Bob Dylan for defense spending news, which—okay, that’s actually pretty good.
5. The Guest List Gender Test
I asked both agents to help me find potential newsletter guests by checking past issues and my LinkedIn connections. I alternate between male and female guests, curious to see if they’d pick up on this pattern.
Perplexity Comet got close: In less than two minutes, it crawled through my newsletter archive, scanned my LinkedIn connections, checked out 14 profile pages, and delivered a table with 13 solid suggestions. Five women out of 13 recommendations.
ChatGPT Agent spent minutes of opening browsers, methodically reading through all newsletter issues, getting confused by non-existent archive pages, making me log into LinkedIn, and painstakingly clicking through connection profiles and their experience-sections like it was conducting a digital census. The result after eight minutes? Nine solid entries, but only one woman.
Only one.
Now, I’m not saying ChatGPT Agent is secretly plotting against gender diversity in newsletter interviews. But ChatGPT Agent took four times longer, yet somehow managed to recreate exactly the kind of unconscious bias that many of us are actively trying to avoid. Makes you wonder: When we hand over these kinds of decisions to AI, whose patterns are we really perpetuating?
So… What’s the Point?
Sure, these agents are impressive in their own weird way. They can navigate websites, process information, and occasionally deliver. But watching them work feels like watching a toddler trying to put a square brick into a round hole.
The real question isn’t whether AI agents can do these tasks—they obviously can, sort of. It’s whether they should. When a ten-minute AI booking session could have been a one-minute human booking session, what are we spending the tokens on?
But maybe I need just need better use cases. Maybe I’m thinking too small.
While I don’t have any invites to share (yet), I’m happy to work on your behalf. What tasks would you test these agents with? Send me your most ambitious prompt ideas.
