OpenAI Launches Dots: Bubbly Background AI Agents That Work On Autopilot
TL;DR
- OpenAI has unveiled Dots, persistent, bubbly avatars designed to work continuously in the background to complete user-defined goals with minimal supervision.
- Dots are hardware-agnostic and live across ChatGPT, mobile, and the web, coordinating tasks across apps, browsing, coding, and messaging on your behalf.
- The launch marks OpenAI's biggest push yet toward fully autonomous personal AI, raising big questions around privacy, control, and the future of work.
Meet Your New Background Workforce
OpenAI just gave personal AI a face — actually, a whole swarm of them.
On Monday, the company unveiled Dots, a new generation of bubbly, animated agentic avatars that live and work quietly in the background, pursuing long-term goals for users without needing constant prompts or oversight.
Described by OpenAI as autonomous companions rather than chatbots, Dots are designed to operate independently of any single device or interface. Whether you're in ChatGPT, on your phone, or away from your keyboard entirely, they keep working — researching, planning, booking, coding, shopping, and coordinating across your apps.
CEO Sam Altman framed the launch as the beginning of the true agentic era: AI that doesn't wait to be asked, but knows what you want and goes and does it.
What Are Dots, Exactly?
Visually, Dots are simple: small, translucent, colorful bubbles that float, bounce, and morph. Each Dot represents a live agent with its own task, context, and personality. You can have one Dot tracking your flights, another negotiating your inbox, and a third building a website — all at the same time.
Under the hood, Dots are powered by OpenAI's latest reasoning models combined with a new persistent agent framework. Unlike traditional ChatGPT sessions that reset when you close the window, Dots have long-term memory, ongoing objectives, and access to a broad toolkit for web browsing, code execution, third-party app actions, voice calls, and file handling.
Tell a Dot something like "plan my Tokyo trip under $2,500 and keep me updated" or "monitor this competitor and brief me every Friday," and it will spin off, break the goal into subtasks, and keep iterating for days or weeks until it's done.
Hardware-Free By Design
The most radical part of Dots is that they aren't tied to hardware.
OpenAI says Dots live in the cloud, not on your phone or laptop. You can start a task in ChatGPT on desktop, check in via voice on mobile, get a ping on WhatsApp or Slack, and let the Dot continue overnight without your device even being on.
The company is positioning Dots as interface-fluid. They can appear as floating overlays in ChatGPT, as widgets on iOS and Android, as voice companions, and eventually via API inside enterprise tools like Salesforce, GitHub, and Notion. Developers will be able to embed Dots into their own apps using the new Dots SDK.
In a demo, OpenAI showed a Dot that watched a user's calendar, email, and travel apps simultaneously, rebooking a canceled flight, rescheduling three meetings, and messaging a hotel — all while the user was asleep.
From Prompting To Delegating
Dots signal a major shift in how we interact with AI: from prompting to delegating.
Instead of back-and-forth chatting, users set intent, constraints, and approval levels, then step back. Dots operate on three autonomy modes: Notify First, Auto-Act with Undo, and Fully Autonomous. Users can pause, rewind, inspect the Dot's step-by-step reasoning trail, or take over at any time.
For consumers, that means AI that can manage job applications, haggle with customer support, monitor prices, or plan events. For businesses, OpenAI is pitching Dots as always-on digital employees for sales research, customer onboarding, QA testing, and operations.
The company confirmed Dots will roll out first to ChatGPT Pro and Business users in the U.S. in the coming weeks, with Plus and free-tier access to follow with usage limits.
Autonomy Comes With Big Questions
As with any leap toward autonomy, Dots arrive with scrutiny.
Privacy advocates are already asking how comfortable users should be with agents that continuously read emails, browse on their behalf, and retain long-term memory across tasks. OpenAI says Dots run with isolated permissions, encrypted memory, on-device confirmations for sensitive actions like purchases and payments, and a full activity log users can delete.
There are also concerns about errors at scale — what happens when an unsupervised agent books the wrong flight, sends the wrong email, or loops through paid actions. OpenAI says guardrails, spending caps, and human-in-the-loop checkpoints are built in by default, but admits the system won't be perfect at launch.
Competitors are watching closely. Google, Anthropic, and Meta have all been racing toward background agents, but OpenAI's bet is that personable, persistent, visual agents people trust will win.
What This Means For The Future
Dots may look playful and bubbly, but the strategy is serious: make AI disappear into the background of life.
If OpenAI gets it right, the next era of computing won't be about opening apps or typing prompts. It will be about managing a fleet of tiny digital workers that handle the boring stuff while you focus — or sleep.
As Altman put it during the announcement: "ChatGPT answered questions. Dots get things done."
The age of autopilot AI has officially begun.
Get All The Latest Updates Delivered Straight To Your Inbox For Free!