OpenAI has recently unveiled its latest innovation, the Dots, during the DevDay 2026 event in San Francisco. These AI agents are designed to be perpetually active, performing tasks on behalf of their users. Dots are characterised by their friendly, rounded shapes, allowing for personalisation and the assignment of multi-step projects to be completed autonomously.
Powered by the advanced GPT-6 Astra model, Dots stand out for their “always-on” functionality. Unlike traditional single-turn chatbots, these agents continuously scour the web, striving to accomplish any tasks you set for them. Furthermore, Dots draw context from connected applications and progressively learn about your preferences over time, enhancing their ability to assist you.
Practical Applications of Dots
In a demonstration shared by OpenAI, a user’s Dot proactively assisted in planning dinner. By analysing the user’s calendar, the Dot realised that the user would be occupied at dinner time and promptly sent a message presenting two food delivery options, complete with pricing details. The user simply indicated their choice and specified when they wanted the order placed. In another scenario, a user collaborated with their Dot to successfully launch a new website.
Users can communicate with their Dots through platforms such as ChatGPT, Slack, and Microsoft Teams, with user context being shared seamlessly across these channels. Additionally, those subscribed to the Pro version can join a waitlist to send text messages to their Dots via iMessage or RCS messaging on Android devices.
Subscription and Features
The rollout of Dots begins today for ChatGPT Pro subscribers, priced at $100 per month. OpenAI typically introduces new features first to this premium subscription tier before expanding access to a broader audience over time. Initially, users will manage a single Dot, but there are plans to enable control over multiple agents in the future.
Dots are designed with user security in mind, requiring explicit authorisation before undertaking sensitive actions such as installing software or changing passwords. Users can activate a “Custom Rules” feature to set clear boundaries that the agents must not cross, as well as tasks that necessitate more direct permission for completion.
Shifting Landscape of AI Agents
The emergence of Dots comes at a pivotal moment for AI agents. Earlier this year, early adopters in Silicon Valley began exploring tools like OpenClaw, while the recent launch of Meta’s Muse agent has attracted a wider user base keen to experiment with personalised agents. Muse quickly ascended the charts of smartphone app downloads, but Dots are poised to challenge Meta’s dominance in this space.
This development is still unfolding, so stay tuned for further updates.