OpenAI's Dots Brings Always-On Agents to Enterprise Work
OpenAI launched Dots, an always-on AI agent, on September 29, 2026. TechCrunch
The release is positioned as an enterprise AI push that takes on Meta. Dots are agents that pursue user goals across apps on their own, which means they keep working in the background without a new prompt for each step. Reuters The emphasis is on persistent execution and finishing work across apps, rather than single-session chat.
OpenAI published its launch announcement on openai.com under the title 'Introducing dots'. The company describes Dots as "remarkably capable, always-on agents built to handle everything." OpenAI
A dot can be created in the ChatGPT desktop app or in ChatGPT on desktop web by following the onboarding prompts. OpenAI Help OpenAI describes a dot as an always-on agent that keeps work moving across tools and projects and that users can keep talking to while it works. OpenAI Developers Setup stays inside ChatGPT. The idea is conversational supervision of parallel work, where you check in by chat while it handles several jobs at once, much like overseeing assistants by message.
Dots features a bubbly, blobby avatar. The name is crowded.
The domain dot.com belongs to Elon Musk's xAI. It currently redirects to the download page for xAI's Grok chatbot app. Whois records show dot.com was transferred in July, as reported on September 29, 2026. The domain dots.com currently belongs to a long-defunct company. TechCrunch
Separately, undated copy accessible on dot.com promotes "Bots like teammates" and a product called Grok Bot. That copy states Grok Bot logs in once and uses users' apps and websites like a user would. It also states Grok Bot is included with eligible Cursor, SuperGrok, or Teams plans. That language sits outside the dated reporting on the redirect and is best read as background until xAI clarifies current packaging.
The broader context here is the shift from prompt-response assistants to delegated, persistent work. For infrastructure and platform teams, the relevant questions are concrete. How is identity delegated across third-party tools. Where are session tokens, the saved logins that let an agent return, stored. What audit trail, or record of actions, covers work done while the user was elsewhere. How are overlapping goals, retries, and failures reconciled.
Looking at what this means for enterprise adoption, always-on changes the trust boundary. A chat assistant that drafts text is easy to contain. An agent that continues to act across ticketing, docs, code, and SaaS admin consoles while the user talks to it needs scoping, rate limits, and clear stop conditions. In my view, persistence only earns production trust if every background step remains inspectable, attributable, and reversible.
Worth flagging for builders is the collision itself. OpenAI shipped Dots without dot.com or dots.com. One is held by a direct competitor in agents, the other by a defunct registrant. That does not block distribution through the ChatGPT desktop footprint, but it complicates discovery, documentation links, and brand control. Watching my own children adopt new tools, the name they remember is the domain they type. Exact-match domains still shape that habit.
Looking ahead, if the operational controls hold, the upside is familiar from prior platform shifts. Long-running, low-attention work moves off human queues. People supervise outcomes rather than clicks. That is the promise OpenAI is selling with Dots, and the reason Meta and xAI are racing on parallel agent tracks.
