OpenAI's dots: how a coding tool became a $100-a-month AI assistant

OpenAI released dots, an always-on AI personal assistant, at its developer day in late September, and at TechCrunch Disrupt this month the company explained the product's origin: it grew out of the coding tools OpenAI built for software engineers.

Alexander Embiricos, who leads OpenAI's enterprise product team, said the assistant represents a deliberate reframing of technology OpenAI already had. The stakes are significant. The always-on assistant — one that keeps working even when nobody is asking it to — has become this year's answer to how people will actually use deep learning models. The startup Instinct became an overnight sensation with the concept, and Meta put its full marketing force behind its own assistant, Muse. OpenAI was not going to be left out.

Why an 'entity' instead of a chat window?

The industry's turn toward cutesy mascots as avatars for trillion-parameter models has raised eyebrows, but Embiricos said the framing is functional, not decorative.

"There's kind of this convergent evolution happening," Embiricos said of the efforts to build AI assistants. "As we were working on the product, we realized if we could frame this as, it's an entity and it's a single conversation, not many different conversations….it starts to become intuitive."

That design decision governs everything about dots. Users get one persistent thread rather than a series of disconnected chats, and the assistant treats itself as a continuous presence rather than a tool summoned on demand.

The name itself is a play on the negative space inside OpenAI's logo, and it is intended "to feel like just something that could eventually be ubiquitous." Not every asset cooperated with that ambition. Asked when he learned that rival AI entrepreneur Elon Musk owns the dot.com URL, Embiricos said: "I found out very late."

What does dots cost, and what does it actually do?

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 a user's calendar, email, computer and other tools, the agent tries to help get work done.

Much of that functionality existed before dots. Embiricos pointed out that users could achieve a similar result with OpenAI's Codex app, its coding tool released at the beginning of this year — but that requires more work and hand-holding. He recalled celebrating when his wife started leaving her computer on all night so that her agents could keep working.

"On the product team, what I think about is, how do we make this intuitive and useful to people?" he told Disrupt attendees. "A lot of that isn't building some new piece of tech into the product; it's how you frame it to the user and how it feels."

The persistent-agent pitch is that the assistant does not need prompting. For OpenAI, that means more token usage. For users, it means a more proactive assistant "that's sort of thinking about you and thinking how to help you," as Embiricos put it.

How well does the proactive part work?

In testing, the assistant's ability to think ahead is apparent and often useful, but it is not quite there yet. It will flag calendar conflicts and upcoming deadlines, but not necessarily in a timely fashion.

Embiricos offered his own experience as evidence. "The first time it told me about one of those [conflicts]—this is like early in development—it told me three minutes before the thing," he said of his dot. "It's like, 'Hey, this thing is coming up. You might want to do something about it.' I only have three minutes!"

He expects the timing to improve over time, but the proactive approach is the core of the product. He pointed to a story from an engineer at a drone delivery startup: "We had this outage, and by the time I noticed that the outage was going on, my dot had already noticed the outage was happening, had already analyzed the outage, and had come to me with useful information for how we could resolve the outage."

How much personal data does dots collect?

As with most AI agent use cases, there is a tension between the personal information an AI model needs to be useful and how much personal data a user exposes to a software company. Meta's Muse has been criticized for the profiles it builds of users' family and friends — profiles that let the software know what kind of restaurant a user might want with a specific friend, but that could also add value to Meta's advertising business.

dots does something similar. "Yes, we have a thing that we built into dots, which we call proactive research," Embiricos said. "On a cadence, it will think about what it knows about you, what is the conversation that it's had, and then what have you given it access to that it can think about, and then it will go and research, 'What could I do to be more helpful to you?' And then, basically, it will propose that."

OpenAI says the data needs of the agent are difficult to separate from those of its parent corporation. Hark, a startup developing its own AI personal assistant, is explicit on this point: it does not collect data to fuel an advertising business the way OpenAI and Meta do.

With dots, OpenAI says the agent carries the same privacy controls as ChatGPT itself:

• Users can turn off model training, stopping OpenAI from using their conversations to improve its AI.
• Users can control what resources the agent accesses on their computer, and how.

Those controls are not granular, and the app can be pushy. It refuses to accept read-only access to cloud drives, for example.

"What I think we need to do is basically build out super, super, super granular controls," Embiricos said. He pointed to the model's intelligence and discretion as the other side of the coin, and to the company's deliberate choice at launch.

"This is actually something we kind of got some flack about when we launched," he said. "We had really conservative defaults over what the agent would do, and it would ask for permission a lot. I remember knowing that we were going to see that feedback, but thinking, 'This is okay, we can tune that over time.'"

Can personal AI agents scale?

The other big question about AI agents is whether they can scale. They work when only a few people have them — stories of agents snagging hard-to-get reservations abound — but businesses do not necessarily like dealing with agents and might not be equipped to.

People do not like dealing with them either. Embiricos counts himself among that group, and he also employs a human assistant, because OpenAI's corporate partners want to interact with a person. "I personally never intend to have a meeting with someone else's agent," he said.

If the AI industry is right that agents are the future, the open question is whether that future arrives without overwhelming the infrastructure the world currently runs on — businesses, websites, payment networks.

"At OpenAI…we're not always moving the fastest in terms of shipping the next ability for the agent to work with the world," Embiricos said. "When we do, we're doing it really closely in collaboration with the rest of the ecosystem. I probably can't say too much here, but we're deep in lots of conversations with payment processors."

Letting an AI assistant spend money through payment-company partnerships would be a potentially lucrative first for OpenAI. It would also be a big ask for consumers: granting an assistant access to an email inbox already stretches comfort, and a bank account remains, for now, off limits.

Embiricos's framing of the launch suggests OpenAI sees the current product as a tunable starting point — conservative defaults, coarse controls, imperfect timing — rather than a finished system, with payment integration and finer permissions as the visible next steps.