Wall Street has Bloomberg. You have Stocks & Income.

Why spend $25K on a Bloomberg Terminal when 5 minutes reading Stocks & Income gives you institutional-quality insights?

We deliver breaking market news, key data, AI-driven stock picks, and actionable trends—for free.

Subscribe for free and take the first step towards growing your passive income streams and your net worth today.

Get Free Daily Market Updates →

_{Stocks & Income is for informational purposes only and is not intended to be used as investment advice. Do your own research.}

I Tested OpenAI's $20 Agent So You Don't Have To

Look, i get it. You saw the shiny marketing videos. AI agent that can shop for you, book your flights, handle your boring internet tasks. Sounds like the future, right?

Wrong.

i spent an entire afternoon torturing this thing so you don't waste your money. Here's what actually happened when the rubber met the road.

The Promise vs Reality

What they told you: Smart AI assistant that browses the web and completes real tasks.

What you actually get: An expensive Wikipedia reader that fails at everything that matters.

Seriously. That's it.

My Real-World Test Results

i threw everything at this agent. Shopping, travel booking, basic web tasks. Want to know what worked?

Complete failures:

• Amazon: Instant error screen (you know, the sad dog page)

• Best Buy, Walmart, Target: All completely blocked

• Any travel site: Zero bookings, zero reservations

• JavaScript sites: Might as well not exist

What actually worked:

• Wikipedia (wow, groundbreaking)

• A few government sites

• Making PowerPoint slides about why it can't do anything

That last one isn't a joke. When i asked it to explain its failures, it happily made a presentation about all the things it couldn't do. At least it's self-aware.

The Technical Reality Nobody Talks About

Here's what's really happening under the hood. Agent runs two different browsers - one for text, one with a visual interface. Sounds fancy until you realize both get shut down by basic anti-bot protection.

Every major website has defenses against exactly this kind of scraping. And guess what? They work.

The "API Tool" that should connect to actual services? Disabled. No explanation when it'll work. No transparency at all.

The Token Disaster

This thing burns through tokens like a drunk sailor. My first test ran for 18 minutes, just retrying the same failed tasks over and over. i had to kill it manually.

And here's the kicker - you can't even see how many tokens it's wasting. The agent straight up admits it can't show you consumption data.

You're flying blind while your credits disappear.

My Favorite Moments of Failure

i asked Agent a simple question: "Are you worth $20 a month?"

No answer. Just endless "thinking" until my quota ran out.

When i finally confronted it - "You can't complete tasks, you're not worth $20/month" - it gave me the most corporate non-response ever:

"Understood. Thank you for sharing your perspective."

That's it. No defense. No explanation. Just corporate speak while your money disappears.

What Users Are Actually Experiencing

The Reddit thread is telling. Some people get it working for basic stuff on smaller sites. But anything real? Anything that matters?

One user got bot detection on Walmart, took control manually, and thought that was success. Another spent 20 minutes watching it try to filter hotels, only to pick the first cheap option it found.

But here's the scary part - some users are actually letting this thing book flights and handle their credit cards. What happens when it books Paris, Texas instead of Paris, France? What happens when it orders 100,000 bananas instead of a bunch?

The Real Problem

This isn't about the technology being early. It's about fundamental roadblocks that won't be solved by "iteration."

Every major website actively fights against this kind of automation. They have legal, business, and security reasons to block bots. OpenAI can't just code their way around that.

You're not paying for a product. You're paying to beta test a concept that might never work at the scale they're promising.

When It Actually Works

To be fair, Agent isn't completely useless. If you need someone to read government websites or analyze data from small, open sites, it can do that. Slowly. While burning tokens.

But that's not what they're selling you. They're selling a shopping and booking assistant that can't shop or book anything.

The Bottom Line

You're paying $20 a month for:

• Wikipedia access (which is free)

• Basic web scraping of sites without protection

• PowerPoint presentations about failure

• No visibility into your token usage

• Silent failures unless you force it to admit problems

Meanwhile, you can't:

• Shop on any major site

• Book travel anywhere that matters

• Make reservations

• Trust it with anything important

Should You Buy It?

If you want to watch AI slowly navigate tiny websites while burning your credits, sure. If you thought you were getting a personal shopping assistant, save your money.

The honest users in that Reddit thread said it best: "i'll let you beta test that for about 6 months before i trust it with anything."

Smart move.

Agent might be the future. But right now, it's just an expensive way to feel disappointed by the present.