OpenAI Dots Review: I Tested the New Always-On AI Agent

I spent a late night testing OpenAI’s new Dots agent after the company’s annual DevDay, where more than 2,500 developers gathered to discuss software and the latest artificial intelligence models. The biggest announcement was Dots, a new always-on AI agent designed to proactively help users complete tasks and make autonomous decisions while browsing the web.

After watching a sometimes messy onstage demo and hearing CEO Sam Altman describe his experience with the agent, I wanted to see what it was like to work with my own Dot. I had spent most of September testing Meta’s Muse, another cute and cuddly AI agent, so I was curious to compare the two user experiences.

My First Night With OpenAI Dots

I named my Dot “Toolie” and customized its appearance to look like a pink frog. As midnight approached and my partner slept nearby, I started whispering sweet nothings to Toolie. I opened the personalized bot through ChatGPT and asked which tasks it handled best.

“What I do best is take tedious and time-consuming tasks and turn them into something usable,” the company responded. Okay, sure. Let’s get to work.

What Can OpenAI’s Agent Do?

I asked Toolie to review my ChatGPT history and identify three tasks I could start immediately. The agent found pending data requests that needed follow-up emails and suggested draft responses.

Toolie also pointed out that my upcoming gaycation was poorly planned and suggested better dinner spots. More time to pack the Speedos. The agent then offered to review my report notes and revise a draft pitch before I submitted it to my editor. I declined because that is my favorite part of the process.

AI Agents Are Moving Beyond Chatbots

Early adopters in Silicon Valley remain obsessed with AI agents, from Instinct and OpenClaw to Google’s CC. But with recent releases such as OpenAI Dots and Meta Muse, these tools are becoming more mainstream.

Chatbots may be fading as the dominant AI experience, at least among developers. Now companies are competing to build the next major agent. They promise that AI agents will simplify everyday life by automating tedious tasks such as making doctor’s appointments, arranging carpools and organizing bills.

OpenAI Dots Pricing and Availability

For now, users can access one Dot, although OpenAI plans to support multiple Dots in the future. At launch, the agents are available only to subscribers of OpenAI’s Pro plan, which starts at $100 per month.

By comparison, Meta Muse is free for anyone who wants to download the app or use the website.

GPT-6 Astra Is OpenAI’s Differentiator

OpenAI leaders believe the quality of Dots, powered by the GPT-6 Astra model, will set the product apart. “We’re starting as a premium product. We use a lot of compute,” Altman said during DevDay.

Dots is currently limited to paid subscribers, but that could change. “Of course you should expect us to provide products for the masses of billions of people,” Altman said.

Why OpenAI and Meta Are Making AI Agents Cute

The companies behind these agents appear to be paying close attention to the cultural backlash against generative AI. Their possible secret weapon is cuteness.

From Muse’s Labubu-inspired style to Dot’s second fluff ball, developers of always-on AI agents are embracing adorable aesthetics as a central part of their branding. I previously spoke with an expert who said this strategy could ease the concerns of users who feel uneasy about anthropomorphic software.

A Familiar, Text Message-Style Experience

The developers behind this new wave of personal AI agents are also using an interface that resembles text messaging and group chats. When you send a task to Toolie or Muse, the agent responds with an emoji—a thumbs-up, heart or something similar—to show that work is underway.