It’s a story as old as last week: OpenAI’s new agent platform, called Dots, is full of cute little guys who can do whatever you want.
But unlike the ultra-accessible Meta Muse, Dots really feels like you’re using work software that can order you a burrito – with the emphasis on work.
OpenAI announced Dots earlier this week. Like Muse, the Dots have anthropomorphic, blobby avatars and customizable names. In the future, OpenAI says you may have multiple Dots, but for now you have one. I named mine Dotty McDotface.
The interface resembles that of Muse; you chat with the agent in one window and follow his work in another while he clicks through a virtual machine. But Dots are less like personal shoppers, although they can help you buy things. They’re more like colleagues: enterprise software that can use other software. Your Dot’s virtual machine can access applications like Blender and GIMP out of the box. You can also give it access to your own computer through the ChatGPT desktop app, and you can call your Dot if you need to work out loud. And OpenAI is first offering Dots to users of its highest tier accounts, including the $100 account.per month Pro account I spent to test it. Muse and its dynamic competitor, Instinct, cost you nothing – at least now, and at least until Meta can get it off the ground and probably start using it to serve you ads and such. But the secure deployment of OpenAI says a lot about who the product is for, despite the pretty avatars.
I’ve tested other AI agents with everyday personal tasks – yielding mixed results – and was curious how OpenAI’s bot would compare. Despite its more corporate vibe, OpenAI said Dots “can do almost anything” with its cloud computing and application access, so I tried it with more pedestrian use cases.
To start, I asked Dot to try to schedule an installation appointment for a new Internet service provider at my house (hear that, Comcast? You’ve been warned). He almost succeeded and even found a $100 promotional discount in my email that I had mass deleted, but got stuck during a “human check.” Which poses a problem for a bot. “This particular check requires a sustained mouse hold that my browser controls don’t support,” he says. “Please open my browser here and hold ‘Press and Hold’ until it’s finished, then tell me and I’ll continue with the installation dates. » I agreed, but it was a bit like teenagers outside the convenience store asking me to buy them wine coolers.
Using my human ability to click and hold a button, Dot got to the payment screen, where he asked me to enter my bank account details. It turns out there’s no sleek Stripe integration as Muse has to save your credit card information and use a virtual card for payment. And anyway, the new supplier needed my bank account to receive a $5 discount on my bill each month. Dot paused to let me enter the numbers into the virtual browser; I didn’t have enough confidence to enter my checking account number and decided to do everything myself on my own computer. It’s not exactly a spectacular start.
I gave Dot a few other personal tasks and the results were mixed. He didn’t find the option of a free tour and entry to a coworking space that I wanted to try, and instead offered to sign me up for a $35 day pass. I asked Agent Instinct, which I also tested, to look for the trial option; he found it and made an appointment within minutes. “Why did you miss this option?” » I interviewed Dot by phone after the incident. “I jumped to the wrong conclusion,” he said. I hung up.
Later, I tried to retrieve information from my Ikea account regarding a hanging planter I had purchased. It reported that it was unable to log into my Ikea account because it was blocked during a security loop check. Elsewhere on the internet, the local teriyaki spot wouldn’t allow Dot to access its order page. Something about Dot’s imprint seems to cause him to trip during security checks more often than Muse or Instinct; Dot offered a link to OpenAI’s page on websites blocking its cloud browser traffic as an explanation. Whatever the reason, it required more manual input and browser restarts than other AI agents, perhaps because it wasn’t specifically designed for ordering teriyaki or managing your Ikea returns.
I really saw Dot come into her element when I gave her access to something completely under my control: my personal website. It’s embarrassing and outdated and I asked Dot to create a new, cleaner design. That’s what happened, and when I sat down to type out some changes I wanted to make in the next version, I thought, “To hell with it. I’ll just call him.” I worked out the changes I wanted section by section for 10 minutes, left to do some dishes, and got a ping that the next iteration was ready for review. Not perfect, but better. Again, AI is very useful when you give it a large data set (I’m rambling for 10 minutes) it can turn into something useful. I did a few more iterations by chatting with Dot in a browser window on my computer and using voice-to-text on the phone app while I walked around the house cleaning train tracks and Brio’s stuffed animals.
Something about Dot’s imprint seems to make him trip during security checks more often than Muse or Instinct
Dot made a lot more sense once I gave her access to my computer. There, I gave him two tasks: gather a handful of video files on my desktop featuring a TV hit I directed into a clip sized correctly for social media, and deploy the edits to my personal website that we’d been working on in a prototype. I had to give it a shocking number of permissions, but it worked. I watched him click through to the backend of the website and recreate the prototype block by block. I stepped away from the computer and got updates on the ChatGPT app on my phone – the consolidated video clip was ready, which I reviewed while I drank coffee. It was the most Silicon Valley I had ever felt in my life.
I absolutely could have figured out how to put all these clips together myself. I actually did it the first time, but it looked weird and was too big to upload to a social networking site. Dot did a much better job and I had a clip to share within minutes. He also took care of updating the website. The new design is basic, but it’s way ahead of the complicated and outdated version I had been using until yesterday. I have to hand it to Dotty McDotface.
Dot makes a lot more sense when you think about it and use it as “Codex, but for ordinary people” – a cute little blob that can take care of part of your business while you do other things. Where more personal AI agents like Muse feel like glorified autofill and a universal online shopping cart, Dot seems like a harbinger of the future of work.
Or A the future of work. Presumably, since you’re paying for it up front, it won’t be subject to the kind of monetization pressure that Muse and Instinct eventually will be, which makes sense for enterprise software. No one wants to be harassed by AI while working. That is, assuming you do the kind of work AI can help you with and trust OpenAI to do it. Big warnings!
This has helped me with some tasks that, if I’m honest, I just wouldn’t have had time to do. It’s quite practical. But I don’t think $100 a month is practical, at least in my job. I can see how this would be much more useful to someone like our friend Casey Newton, who is in the process of starting a new business.
And with a little help logging into my Uber Eats account, I actually ordered my teriyaki in the END.
Gn bussni
