{"id":37098,"date":"2026-10-06T07:42:17","date_gmt":"2026-10-06T06:42:17","guid":{"rendered":"https:\/\/www.digife.it\/?p=37098"},"modified":"2026-10-06T07:42:17","modified_gmt":"2026-10-06T06:42:17","slug":"openai-dots-chatgpt-agents-how-they-work","status":"publish","type":"post","link":"https:\/\/www.digife.it\/en\/openai-dots-chatgpt-agents-how-they-work\/","title":{"rendered":"OpenAI Dots: How ChatGPT&#039;s Always-On Agents Work"},"content":{"rendered":"<p><b>OpenAI introduces Dots to ChatGPT: GPT-6 Astra-based agents with their own cloud computer, access to authorized applications, and the ability to perform ongoing tasks even between conversations.<\/b><\/p>\n<p><span style=\"font-weight: 400;\">OpenAI is adding a new category of agents to ChatGPT called <\/span><a href=\"https:\/\/help.openai.com\/en\/articles\/20001530-getting-started-with-your-dot\" target=\"_blank\" rel=\"noopener\"><b>Dots<\/b><\/a><span style=\"font-weight: 400;\">.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The news appears in the release notes of the <\/span><a href=\"https:\/\/help.openai.com\/en\/articles\/6825453-chatgpt-release-notes?utm_source=chatgpt.com\" target=\"_blank\" rel=\"noopener\"><b>September 29, 2026<\/b><\/a><span style=\"font-weight: 400;\"> and is rolling out progressively to some Pro, Business Premium, and Enterprise users.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The difference compared to a normal conversation with ChatGPT is significant.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">A Dot isn&#039;t just meant to receive a request, produce a response, and finish the job. OpenAI describes it as a <\/span><b>always-on agent<\/b><span style=\"font-weight: 400;\">, that is, an always-available agent who can be entrusted with a continuous objective.<\/span><\/p>\n<h2><b>What are OpenAI Dots?<\/b><\/h2>\n<p><span style=\"font-weight: 400;\">A Dot is a personal AI agent available within ChatGPT.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">According to the official documentation, each Dot has:<\/span><\/p>\n<ul>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">one of its own <\/span><b>cloud computing<\/b><span style=\"font-weight: 400;\">;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">access to applications that the user decides to connect;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">memory of the context useful for work;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">ability to carry out complex activities over time;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">ability to ask for user intervention when a decision is needed.<\/span><\/li>\n<\/ul>\n<p><span style=\"font-weight: 400;\">The system is powered by <\/span><b>GPT-6 Astra<\/b><span style=\"font-weight: 400;\">, the model that OpenAI introduced in September 2026 with advanced capabilities for computer use, browsing, coding, and multi-step work.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The relevant transition is therefore this: from a chat used on demand to an agent who can be assigned responsibility.<\/span><\/p>\n<h2><b>What is the difference between a Dot and a regular chat?<\/b><\/h2>\n<p><span style=\"font-weight: 400;\">In a traditional conversation, the user maintains direct control of the flow.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">He&#039;s asking for something.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">He gets a response.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">If necessary, correct or add information.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">A Dot, on the other hand, is configured with a goal and a certain level of autonomy.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">OpenAI explains that it can understand what steps are needed, use the tools provided, and continue the work, reporting the results back to the user when review or decision-making is needed.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The change therefore mainly concerns the <\/span><b>continuity<\/b><span style=\"font-weight: 400;\">.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">You no longer need to re-type the same task every time you reopen ChatGPT.<\/span><\/p>\n<h2><b>An AI agent can work across multiple applications<\/b><\/h2>\n<p><span style=\"font-weight: 400;\">One of the core elements of Dots is the ability to use connected applications.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">OpenAI describes the Dot as an agent that can work \u201cacross the apps you choose to connect,\u201d that is, through the tools authorized by the user.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">This opens up very different scenarios than simply generating text.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">An agent could, for example, collect information from multiple services, update documents, organize tasks, or prepare the output of a multi-step process.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The important point is that the Dot has its own operating environment in the cloud.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">It therefore does not limit itself to explaining what should be done: it can concretely use tools and applications available within the authorised perimeter.<\/span><\/p>\n<h2><b>The real difference is moving from a task to a responsibility<\/b><\/h2>\n<p><span style=\"font-weight: 400;\">\u201cWrite this email\u201d is a task.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">\u201cManage the follow-up of this project and bring me decisions that require my intervention\u201d is a broader responsibility.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">This is the kind of change that the Dots make most evident.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">A continuous agent can be especially useful when the job does not end with a single output, but requires multiple steps spread over time.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">For a company, activities such as:<\/span><\/p>\n<ul>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">collect information periodically;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">keep a project up to date;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">prepare materials from new information;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">coordinate steps between multiple instruments;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">highlight situations that require a human decision.<\/span><\/li>\n<\/ul>\n<p><span style=\"font-weight: 400;\">Technology alone, however, does not solve the organization of the process.<\/span><\/p>\n<h2><b>Before delegating, it is necessary to define what the agent can do<\/b><\/h2>\n<p><span style=\"font-weight: 400;\">Entrusting responsibility to an autonomous system requires more precise rules than simply asking for a draft.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">A business activity often contains exceptions that people know about but have never formalized.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">A sales manager might know that:<\/span><\/p>\n<ul>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">above a certain amount an authorization is required;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">some customers have different conditions;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">certain promises should not be made;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Some information needs to be verified before submission.<\/span><\/li>\n<\/ul>\n<p><span style=\"font-weight: 400;\">For an experienced person these rules may be implicit.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">For an agent they must become readable.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Before delegating a process it is therefore useful to establish:<\/span><\/p>\n<ul>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">what information you can consult;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">what actions it can perform autonomously;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">when you need to ask for confirmation;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">which cases must always be handled by one person;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">what \u201cjob completed\u201d actually means.<\/span><\/li>\n<\/ul>\n<h2><b>AI agents are also changing the way we measure time savings.<\/b><\/h2>\n<p><span style=\"font-weight: 400;\">With a chatbot the result is immediately visible.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">With a continuous agent the evaluation must be broader.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">It is not enough to measure how quickly it produces a response.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">It should be noted:<\/span><\/p>\n<ul>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">how many manual tasks it eliminates;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">how many checks remain necessary;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">how many errors or corrective interventions it introduces;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">how long does supervision take;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">if the end result really comes faster.<\/span><\/li>\n<\/ul>\n<p><span style=\"font-weight: 400;\">An agent that produces a lot of work but takes just as long to control is not necessarily improving the process.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The correct parameter remains the time needed to reach an approved result.<\/span><\/p>\n<h2><b>Are OpenAI Dots available in Italy?<\/b><\/h2>\n<p><span style=\"font-weight: 400;\">To the <\/span><b>October 1, 2026<\/b><span style=\"font-weight: 400;\">, availability is still limited.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">OpenAI says Dots is rolling out to users <\/span><b>Pro<\/b><span style=\"font-weight: 400;\">, but at launch Pro access <\/span><b>does not include the European Economic Area, Switzerland and the United Kingdom<\/b><span style=\"font-weight: 400;\">.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">This means that a Pro user in Italy is currently not included in the initial distribution.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The situation is different for <\/span><b>Business Premium<\/b><span style=\"font-weight: 400;\">, where OpenAI indicates availability in supported ChatGPT regions, and for <\/span><b>Enterprise<\/b><span style=\"font-weight: 400;\">, where a beta phase is planned that can be activated by workspace administrators.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Availability therefore depends on the plan, geography and workspace configuration.<\/span><\/p>\n<h2><b>OpenAI Dots and the Growth of AI Agents<\/b><\/h2>\n<p><span style=\"font-weight: 400;\">Dots fits into a broader market direction.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">OpenAI also introduced new agent-dedicated infrastructures in the same period, including the <\/span><b>Agents API<\/b><span style=\"font-weight: 400;\">, announced on September 10, 2026.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Change is no longer just about models capable of generating better content.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The competition is shifting towards systems capable of:<\/span><\/p>\n<ul>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">use tools;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">move between applications;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">maintain context;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">perform multiple steps;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">work with greater autonomy.<\/span><\/li>\n<\/ul>\n<p><span style=\"font-weight: 400;\">For companies, this means starting to evaluate AI not just as a production tool, but as a potential component of operational processes.<\/span><\/p>\n<h2><b>What this means for your business<\/b><\/h2>\n<p><span style=\"font-weight: 400;\">Before introducing an AI agent, it is a good idea to identify the tasks that currently require continuous manual steps.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">For example:<\/span><\/p>\n<ul>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">open multiple tools to complete a single task;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">copy information from one system to another;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">repeat the same checks over and over again;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">periodically update documents or reports;<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">gather information before making a decision.<\/span><\/li>\n<\/ul>\n<p><span style=\"font-weight: 400;\">From there, you can understand whether you really need an agent, simpler automation, or better integration between the software you already use.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The goal shouldn&#039;t be \u201cuse an AI agent.\u201d.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">It should reduce unnecessary steps while maintaining control, auditability and accountability over decisions.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">DigiFe develops<\/span><a href=\"https:\/\/www.digife.it\/en\/ai-solutions-for-business\/\"> <span style=\"font-weight: 400;\">AI solutions for business<\/span><\/a><span style=\"font-weight: 400;\"> starting from the company&#039;s real processes, identifying where agents, automations and integrations can create a concrete operational advantage.<\/span><\/p>","protected":false},"excerpt":{"rendered":"<p>OpenAI introduce i Dots in ChatGPT: agenti basati su GPT-6 Astra con un proprio computer cloud, accesso alle applicazioni autorizzate e la possibilit\u00e0 di portare avanti attivit\u00e0 continuative anche tra&#8230;<\/p>","protected":false},"author":4,"featured_media":37099,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1,59],"tags":[],"class_list":["post-37098","post","type-post","status-publish","format-standard","has-post-thumbnail","category-curiosita-web","category-notizie"],"_links":{"self":[{"href":"https:\/\/www.digife.it\/en\/wp-json\/wp\/v2\/posts\/37098","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.digife.it\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.digife.it\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.digife.it\/en\/wp-json\/wp\/v2\/users\/4"}],"replies":[{"embeddable":true,"href":"https:\/\/www.digife.it\/en\/wp-json\/wp\/v2\/comments?post=37098"}],"version-history":[{"count":6,"href":"https:\/\/www.digife.it\/en\/wp-json\/wp\/v2\/posts\/37098\/revisions"}],"predecessor-version":[{"id":37144,"href":"https:\/\/www.digife.it\/en\/wp-json\/wp\/v2\/posts\/37098\/revisions\/37144"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.digife.it\/en\/wp-json\/wp\/v2\/media\/37099"}],"wp:attachment":[{"href":"https:\/\/www.digife.it\/en\/wp-json\/wp\/v2\/media?parent=37098"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.digife.it\/en\/wp-json\/wp\/v2\/categories?post=37098"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.digife.it\/en\/wp-json\/wp\/v2\/tags?post=37098"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}