OpenAI’s dots, released at its developer day in late September, is the company’s entry into the current wave of always-on AI personal assistants. The startup Instinct became an overnight sensation, and Meta put its marketing weight behind Muse. Alexander Embiricos, who leads OpenAI’s enterprise product team, said at TechCrunch Disrupt that the shift toward mascot-like agents is a case of convergent evolution. The key design choice was to frame the product as a single entity and a single conversation, rather than many separate chats, so the interaction becomes more intuitive. The assistant is available in the ChatGPT desktop and mobile apps to subscribers on the Pro tier or higher, with plans starting at $100 a month. Once connected to calendar, email, computer, and other tools, it works in the background and tries to help get work done. Embiricos said much of the same functionality is possible with OpenAI’s Codex coding app, but dots requires less hand-holding. He described celebrating when his wife started leaving her computer on overnight so her agents could keep working.

The product is essentially a repackaging of the company’s model capabilities into a more accessible form, following the path OpenAI took when it spun ChatGPT Work out of agentic coding tools. The name dots comes from the negative space inside the OpenAI logo and is meant to feel like something that could eventually be ubiquitous; Embiricos said he found out very late that Elon Musk owns the dot.com URL. The persistent-agent idea means users do not need to prompt it. For OpenAI, that means more token usage; for users, a more proactive assistant that is thinking about them. The article’s own testing found the assistant’s ability to think ahead is apparent and often useful but not quite there yet. It flags calendar conflicts and upcoming deadlines, but not always in a timely way. Embiricos recalled an early development moment when his dot warned him about a conflict three minutes before it happened. He expects the timing to improve, but the proactive approach is central to the product. He also told a story about an engineer at a drone delivery startup whose dot noticed an outage, analyzed it, and came to them with useful resolution information before the engineer had even noticed.

The main privacy concern is the tension between the personal data an agent needs and what a software company collects. dots includes a feature OpenAI calls proactive research: on a cadence, the agent considers what it knows about the user, the conversation history, and what resources the user has granted access to, then researches how it could be more helpful and proposes actions. The article notes that rival startup Hark explicitly says it does not collect data for an advertising business the way OpenAI and Meta do. OpenAI says dots has the same privacy controls as ChatGPT: users can turn off model training and control which resources the agent can access on their computer and how. But the controls are not granular, and the app can be pushy; for example, it refuses to accept read-only access to cloud drives. Embiricos acknowledged the need for super granular controls. He also said the initial version received criticism for conservative defaults and frequent permission requests, and that OpenAI knew it would tune that over time.

A remaining question is whether personal AI agents can scale. They are fun when few people have them, but businesses may not be equipped to deal with agents, and people may not like interacting with them. Embiricos himself has a human assistant because OpenAI’s corporate partners want to talk to a person, and he said he personally never intends to have a meeting with someone else’s agent. On the infrastructure side, OpenAI is not always moving fastest to give agents new abilities to interact with the world, but when it does, it collaborates with the ecosystem. Embiricos said OpenAI is deep in conversations with payment processors, though he could not say much more. Letting an AI assistant spend money would be a potentially lucrative first for OpenAI but a big ask for consumers; even giving an assistant access to an email inbox already feels like a big ask, and bank accounts are off limits for now.