I Gave an AI $100 and Told It to Trade Stocks While I Sleep

Disclosure: This article was drafted with AI assistance and reviewed by the author.

It is about 1 a.m. on a Sunday night, and I just did something that either makes me an early adopter or an idiot. I connected my Robinhood account to an AI trading agent, funded a separate account with $100 of my own real money, and told it to trade while I sleep.

This is part one of a series. I am going to document the whole experiment honestly: the setup, the rules, the wins, the losses, all of it. If it blows up the $100, you will read about that too.

Why I did this

Robinhood recently launched something called agentic trading. It is an official connection that lets an AI agent see your account, read live market data, and place real trades on your behalf. Not paper trading. Real money, real orders, real consequences.

I have always been curious about day trading, but I have never had the time, the knowledge, or honestly the nerve to sit and watch charts all day. The idea that an agent could do the watching, the researching, and the trading around the clock sounded like science fiction. So I decided to test it with an amount I can afford to lose: $100.

That number matters. One hundred dollars is real enough that I care, and small enough that losing it will not ruin my week. It is the perfect size for an experiment.

The rules I set

I did not just hand over the keys with no conditions. Before the agent placed a single order, I made it write down the rules and I approved them. Here they are:

1. Separate account only. The agent trades in a dedicated account with just the $100. My main investment account is completely off-limits. It cannot see it, touch it, or sneeze near it.

2. The $60 kill switch. If the account value drops below $60, the agent stops everything immediately and tells me. No exceptions, no "just one more trade to win it back." This is the single most important rule.

3. Think before every trade. The agent has to write down its reasoning before every single order: why this stock or crypto, what the catalyst is, and exactly what price would prove it wrong. No reasoning, no trade.

4. Daily report. Every weekday afternoon after the market closes, I get a summary: account value, what it bought and sold, and how close we are to the goal.

5. The goal. Turn $100 into $150 in one week. Ambitious? Yes. That is a 50 percent gain. But this is an experiment in aggressive growth, and I told the agent to hunt for real opportunities, not to force trades when the market is flat.

Night one: nothing happened, and that is the point

Here is the honest part. As I write this, it is around 1 a.m. The agent has been watching the markets for a while now. It checked crypto prices. It checked a few volatile stocks. And it placed exactly zero trades.

Nothing was moving with enough conviction. Crypto was flat. The stocks on its watchlist were drifting a percent or two with no catalyst behind the moves. So it sat out.

A human trader, me included, would have probably forced a trade out of boredom. That is how small accounts die: death by a thousand bored trades, each one bleeding a little on the spread. The agent looked at a flat market and did the hardest thing in trading, which is nothing.

I know "nothing happened" is not the exciting update anyone wants on night one. But discipline is the entire thesis of this experiment. If the agent only trades when there is a genuine setup, the $100 has a fighting chance. If it traded for entertainment, the $100 would already be shrinking.

What happens next

The agent runs on a schedule now. Every 30 minutes overnight, it watches crypto. Every 10 minutes during market hours, it scans stocks: live prices, breaking news, chart patterns, momentum. When something genuinely ignites, it can act in minutes. I will get a message for every trade it makes and a full summary every afternoon.

I have no idea how this ends. Maybe the agent strings together a few good trades and we hit $150 by Friday. Maybe the market chops sideways all week and we end at $102. Maybe we hit the $60 kill switch and I write a very different article next week.

That uncertainty is the whole reason I am writing this series. Everyone talks about AI in theory. I want to see what it does with real money, real markets, and real rules, and I want you to see it too.

Check back for part two. The experiment is live.