OpenAI Dot user reports failed calls days after agent launch
The user said calls to their Dot failed across multiple phones, even though it had completed a 28-minute call the day before.
By Ryan Merket · Published
Primary source: X
Why it matters
Dots are designed to stay available between chats, but a user reports that one of their supported contact methods failed. The post points to a possible fix while illustrating how aggregate service status can miss an individual access problem.
A user of OpenAI's new Dots agent reported on October 1st that calls to their Dot were failing across every phone they tried, two days after OpenAI introduced the always-on assistant. The user, Developing Adventures (@DevAdventur3s), said the same Dot had completed a 28-minute call the previous day and displayed a "Call failed" message on subsequent attempts.
The post describes one user's problem, not a confirmed service-wide outage. Developing Adventures said OpenAI's status page showed Dots at 100% availability and no incident. The post also says OpenAI product leader Thibault "Tibo" Sottiaux (@thsottiaux) had said a fix was coming and suggested a Dot might be busy when it did not answer. That account of Sottiaux's response comes from the user post; it does not establish how many users were affected or when a fix would ship.
The report lands early in Dots' rollout. OpenAI announced the product at DevDay on September 29th, describing Dots as agents powered by GPT-6 Astra that can keep working on goals between conversations, use their own cloud computer and browser, and connect to apps. OpenAI says users can reach a Dot through ChatGPT, Slack or Teams, and can also call one through ChatGPT. The call failure described here concerns that conversational access, rather than a claim that the Dot failed while carrying out a background task.
OpenAI's help documentation draws another boundary: a Dot cannot initiate calls to its user at launch. Users can call their Dot, but the product is not set up to ring them independently. The distinction matters when judging the report: it describes an expected way of contacting the agent failing for one user, not a Dot failing to alert its owner.
Access is also limited during rollout. OpenAI says Dots are being made available gradually to eligible Pro and Business Premium users in supported markets; Enterprise access is a beta that workspace administrators must enable. The company says the first Dot is included with eligible plans at no extra charge, while access may take several days to reach an account. These limits mean the report cannot establish how widely the feature is being used or whether the failure is tied to a particular account, device, carrier or region.
OpenAI's public status page reports service availability at an aggregate level across tiers and features, and warns that individual customer availability can vary. A green system indicator therefore does not rule out a problem affecting a particular user's calls. The status snapshot in the post is useful context, but it does not diagnose the cause.
OpenAI is positioning Dots as persistent agents that remain available beyond a single chat session. In that product design, voice is one route for users to reach an agent whose work may continue in the background. A failed call does not show whether the Dot's ongoing tasks were affected, and the post supplies no evidence of a broader reliability problem. It does expose a basic launch-day test: a user needs to be able to reach the agent through the channels OpenAI says it supports.
The incident is small in scale and specific in scope. Its significance will depend on whether the fault was isolated or recurs as access expands. For now, the evidence is a single user's report, their description of Sottiaux's response, and OpenAI's stated product design and rollout terms.