I used OpenAI's Dots to help me move offices. It spammed the movers with hundreds of emails.

• Robert Farlow asked OpenAI's Dot agent to help coordinate a move.

• His Dot spammed the movers, building managers, and insurance workers with hundreds of emails.

• He shut the Dot down, though AI is still writing his emails.

This as-told-to essay is based on a conversation with Robert Farlow, the 27-year-old founder of the AI startup Plato. He is based in San Francisco. This essay has been edited for length and clarity. OpenAI did not respond to Business Insider's request for comment.

We just raised a new fundraising round, and we're growing the team a lot. So we're getting a new, bigger office .

It's such a headache. There's so much insurance stuff and coordinating with the movers. The building managers require insurance when the movers come in. We also had to get new insurance to ensure we were in compliance with the new lease.

I use all of the AI agents. Instinct helps me at the gym, and Grok Bot takes snack orders at my company and buys them on Amazon. I would've had to ask Claude or ChatGPT about the move anyway, because I don't know much about insurance.

So, I thought I'd give this task to Dot and let it take it from there.

I had found a contact for insurance. I said: "We're moving to this office. You can read my emails to find out the details. Just get it done." A couple of times it sent me a bunch of texts. I said: "Just please, get it done."

It was taking a few days, and the movers still hadn't been finalized, so I looked at what it was doing.

It was mayhem.

It was sending multiple back-to-back messages to the movers, triple- and quadruple-emailing, giving every tiny update about everything.

They proposed some times and said, "How's 9 a.m.?" It followed up: "Is that Pacific Time?" It needed clarification on every single thing, so it just kept bombarding them with questions.

There were also the insurance people , our current building managers, and our new building managers. In some cases, it was emailing all of them at the same time.

The new building manager had to make a new thread because the existing one was too large.

It was telling them stuff about the other parties, too. It was telling the movers about Brian, our building manager. They don't need to know that. It was giving them way too much information.

To all parties, it probably sent around 300 emails in a few days.

At first, they didn't know it was AI. Clearly, the movers caught on. I think the other parties never found out; they probably just thought it was some annoying guy spamming them with a bunch of other stuff.

I saw the email from the movers and felt so bad. I'm like, "What did I do?" I shut the Dot down right after I saw it. I haven't responded yet, but I want to respond because I put them through a lot.

It didn't work out with those movers anyways, because they didn't have the necessary insurance. I got some other movers, and I didn't use Dot this time. I still use AI to draft, but I review it before sending.

The models will get better at this, and I think you're going to see more of this. For now, unless you're absolutely sure the models will do a good job, you should review everything your Dot does.

But I also think this is a glimpse of what the future will look like. Soon, it'll be another agent, not a person, responding to all this.

If you enjoyed this story, be sure to follow Business Insider on Yahoo.