OpenAI launches Dots, an always-on AI agent with an animated avatar

OpenAI launches Dots

On Tuesday, OpenAI introduced Dots, a new product that TechCrunch describes as an always-on AI agent, presented to users through a playful, blob-like animated character rather than a conventional chat window. Beyond the mascot, the announcement — as summarized in a developer-focused article on dev.to — positions Dots as assistants that take in feedback, connect to other apps, and pursue a user's goals over time, instead of trading single prompts for single answers.

That shift from conversation to ongoing agency is what makes the avatar more than decoration. The dev.to piece, by character designer and Rive animator Praneeth Kawya Thathsara, is an independent design discussion (the author notes it does not describe any custom integration with Dots), but it frames the problem such an interface has to solve: an agent working in the background can be listening, processing, working, speaking, waiting for approval, reporting completion, or explaining a failure — and users need to tell which is which at a glance.

The character as a status signal

Thathsara's argument is that an expressive character can carry much of that state information alongside text. He maps application states to character behavior — an idle companion breathes and blinks, a listening one looks attentive, a working one settles into a restrained activity loop — while keeping the application itself the source of truth, so a completion animation only fires when a task is genuinely finished.

He also recommends separating three control channels: activity, emotion and speech, so that a concerned expression can persist while the character talks. Interactive animation tools such as Rive, he writes, let developers drive these states from real application events rather than playing back fixed video sequences.

dot.com leads to Grok

The launch also produced a subplot that TechCrunch covered the same day. The domain dot.com is registered to Elon Musk's xAI and currently redirects visitors to the download page for the Grok chatbot app. Whois records reportedly show the domain changed hands in July, and an anonymous X user who tracks xAI, @birdabo, surfaced the redirect in a post that spread widely.

There is an innocent reading: "dot" could be a mistyped "bot", so xAI might have bought the address to capture typos. TechCrunch, which says it has contacted xAI for comment, points out the weak spot in that theory — xAI does not own bot.com, and another obvious typo domain, vot.com, is simply listed for sale. The more popular theory online is that someone at xAI got wind of the product and its name and grabbed the domain as a jab at OpenAI — the company Musk left before launching the competing Grok chatbot and filing an unsuccessful lawsuit against his former employer. Meanwhile, dots.com, the closer match to the product's actual name, belongs to a long-defunct company, as TechCrunch notes.

Why it matters

Dots is a marker of where consumer AI interfaces are heading. Persistent agents that act across apps need to communicate their state continuously, and OpenAI is evidently betting that an animated character is part of the answer. The design questions Thathsara raises — what the face does while the agent works, waits or fails — will apply to any developer building similar products, and Dots gives them a high-profile reference point.

The domain episode is partly comedy, but it is also a reminder of the low-grade hostilities between OpenAI and Musk's xAI, and a practical lesson in product naming: launch a flagship product without the matching web address, and someone else may already be holding it.