OpenAI Unveils Dots as Always-On Agents Try to Close the Gap With Meta

OpenAI launched Dots at DevDay 2026, introducing always-on agents designed to pursue user-defined goals across connected applications. The release addresses a competitive gap around persistent consumer assistants, a category where Meta’s Muse has attracted attention before OpenAI’s entry.

Dots represent a shift from the familiar chatbot session toward background software that keeps working after a user leaves the interface. TechCrunch described the product as a personal agentic assistant powered by GPT-6 Astra, while The Verge framed the launch as OpenAI’s response to a market that has moved beyond general-purpose APIs. The company presented more than 20 launches at its San Francisco developer event, according to The Verge’s event coverage.

Dots are built to operate independently of a specific device or interface, with access to connected apps and the ability to handle multistep tasks in the background. OpenAI describes the agents as capable of handling nearly anything, while coverage emphasizes their persistent operation and minimal need for oversight.

The product’s presentation also gives it a distinct consumer identity. Wired reported that Dots are represented as cute blobs, a design choice that puts a friendly face on software with permission to act across a user’s digital life. That framing matters because always-on agents require repeated access to applications, personal information, and ongoing tasks. The Verge and TechCrunch both connect Dots to the broader push for assistants that pursue goals continuously rather than answer isolated prompts.

OpenAI’s challenge is strategic as much as technical: the company must turn its strength in language models and developer infrastructure into an agent product people trust to act without constant supervision. The Verge’s pre-event analysis said OpenAI had fallen behind in continuously running, consumer-facing agents, while Finance.yahoo described DevDay as a test of whether the company would pivot from the general-purpose API market.

Meta’s Muse provides the clearest rival named in the coverage. OpenAI’s Dots launch places both companies in a race to define the everyday assistant, with persistence, app connections, and initiative becoming central product features. Bloomberg also positioned Dots against Meta and described a new $500 paid tier, indicating that OpenAI is pairing the consumer agent push with a premium monetization strategy.

OpenAI also introduced AgentKit at DevDay, signaling that the company is targeting builders alongside end users. The available OpenAI announcement identifies AgentKit as a product for developers, while Forbes reported plans for managed agents at the event. Together, the announcements place Dots inside a broader effort to make agent creation and operation part of OpenAI’s platform business.

That platform emphasis reflects a wider change in agent development. An n8n analysis published in April described major technology companies entering a market shaped by tools such as OpenClaw and security standards around the Model Context Protocol. OpenAI’s developer programming, including DevDay Exchange events planned for Bengaluru, Tokyo, and other cities, gives the company a distribution channel for turning experimental agents into deployable applications. The technical value of AgentKit will depend on controls, integrations, reliability, and operating costs, areas the launch coverage does not fully detail.

Dots’ success will depend on whether users trust an always-on agent to act correctly, protect sensitive data, and stop when a task crosses a boundary. The product’s promise of background work raises a higher bar than ordinary chat, because errors can affect calendars, purchases, files, messages, and other connected services without an active prompt in front of the user.

OpenAI’s event strategy puts that trust question alongside a large launch slate. CNBC’s live coverage described Dots as one of the products revealed as protestors rallied outside the venue, while Bloomberg highlighted the product’s proactive design and premium price point. OpenAI’s own DevDay materials focused on technical sessions and new tools for developers, but the public product framing is centered on delegation. For builders, the practical test will be whether AgentKit provides clear permissions, observability, testing, and recovery when agents fail.

OpenAI now has to prove that Dots are more than a high-profile answer to Meta’s Muse. The immediate questions concern availability, pricing, supported applications, model limits, and the safeguards governing continuous activity. Bloomberg’s report of a $500 paid tier makes cost part of that evaluation, especially if the product targets professionals who expect measurable value from autonomous work.

The company’s developer outreach gives it a route to expand beyond a single assistant. AgentKit can help OpenAI recruit businesses and independent developers, while global DevDay Exchange events extend that effort into regional builder communities. OpenAI’s launch therefore combines a consumer-facing agent, a managed development platform, and a premium commercial offer. The outcome will depend on execution: reliable task completion and transparent controls will matter more than animated avatars or the size of the launch list.