Agentic AI · BraivIQ AI Blog

The Agent That Never Clocks Off: OpenAI's Dots Are Always-On AI Coworkers - And UK Pro Users Can't Have Them Yet

At its 2026 DevDay, OpenAI made the clearest statement yet about where AI is going: away from a chat box you visit and towards a colleague that keeps working when you have closed the tab. Dots are persistent, always-on agents powered by GPT-6 Astra. Each runs in its own cloud environment with a browser, connects to more than 4,000 apps, works through Slack and Teams, takes goals rather than one-off prompts, and learns your preferences over time - updating sales proposals, researching, analysing data, preparing documents and building software while you get on with something else. They went live on 29 September. There is a catch for British readers: the Pro rollout excludes the UK, the EEA and Switzerland on regulatory grounds. This featured analysis explains what Dots actually are, why the shift from asking AI to delegating to it is the most important change of the year, where always-on agents genuinely help and where they need tight governance, what the UK exclusion really signals - and how a UK business can build the same always-on pattern today on platforms it controls.

October 3, 2026 · 12 min read · By BraivIQ Editorial

4,000+ apps - Each Dot can connect to more than 4,000 applications, including Slack and Teams · Always on - Dots keep working between conversations - you give them goals, not one-off prompts · 29 Sept - 2026 - Dots went live, powered by GPT-6 Astra with dedicated cloud compute · Not UK (yet) - The Pro rollout excludes the UK, EEA and Switzerland on regulatory grounds

For three years, using AI has meant visiting it. You open a chat window, ask for something, get an answer, and the AI waits - inert - until you come back. At its 2026 DevDay, OpenAI made the clearest statement yet that this model is ending. Its headline launch, Dots, are persistent agents that do not wait. Each Dot is powered by GPT-6 Astra and runs in its own cloud environment with its own browser and plugins, connects to more than 4,000 apps, works where your team already works - Slack, Teams - and, crucially, keeps going between conversations. You do not prompt a Dot. You give it a goal. It manages an evolving project, keeps a sales proposal current, researches a market, analyses incoming data, prepares documents, even builds software, and it learns from your feedback what you want over time. They went live on 29 September. There is one detail every British business owner should notice: the Pro rollout excludes the UK, the EEA and Switzerland, citing regulatory constraints. As an AI Agency London that builds always-on agents for UK clients, we think Dots are the most important product signal of the year - and the UK exclusion makes understanding them more urgent, not less. This featured analysis explains what they are, what they mean, and what to do about them.

From Asking To Delegating - The Shift In One Idea: A chat assistant is a tool you use: you are present for every step, and the AI does nothing when you leave. An always-on agent is a colleague you delegate to: you set the objective once - 'keep our competitor pricing sheet up to date every week', 'chase overdue proposals and draft follow-ups', 'monitor these accounts and flag anything unusual' - and it works continuously in the background, coming back to you when it has something or needs a decision. The difference is not intelligence. It is time and initiative. Delegation is how organisations actually get leverage from people, and Dots are OpenAI making it how they get leverage from AI. Once you see it, the question for every business changes from 'what can we ask AI?' to 'what recurring work can we hand to an agent that never clocks off?'.

Where Always-On Agents Genuinely Help

The value of an always-on agent is concentrated in a specific kind of work: recurring, multi-step, spread across several tools, and currently done by a capable person who would rather be doing something else. That describes a surprising amount of every business. The weekly report that someone compiles by pulling numbers from three systems. The pipeline review where someone checks every open deal for staleness and drafts the nudges. The market and competitor watch nobody has time to do properly. The inbox triage that sorts, summarises and routes before anyone reads it. The project that needs someone to keep the plan, the documents and the status in sync. None of these needs genius. All of them need persistence, and persistence is exactly what a chat assistant lacks and an always-on agent provides. The learning element matters too: an agent that remembers how you like proposals structured, which accounts matter most and what 'unusual' means for your business gets steadily more useful, rather than starting from zero every morning. For a growing business, that is the closest thing yet to adding capacity without adding headcount.

Recurring reporting - pulling figures from several systems on a schedule and producing the summary nobody enjoys writing.
Pipeline and account upkeep - checking for stale deals, drafting follow-ups, keeping proposals current.
Monitoring - watching competitors, markets, accounts or data feeds and flagging what matters.
Inbox and request triage - sorting, summarising and routing before a person needs to read anything.
Project coordination - keeping plans, documents and status in sync across the tools a team already uses.

Where They Need Tight Governance

The same properties that make always-on agents valuable make them risky, and a business should be clear-eyed about it. An agent that can connect to 4,000 apps has a correspondingly large blast radius: every connection is a system it can read and, potentially, act in. An agent that works while you are not watching can make a mistake that runs for days before anyone notices. An agent that reads email and web pages can be manipulated by instructions hidden in that content. None of this is a reason to avoid always-on agents - it is a reason to deploy them deliberately. Grant each agent access only to the specific systems its goal requires, not everything it could connect to. Decide in advance which actions it may take alone and which need a person's approval: drafting a follow-up is fine unattended. Sending a price change to a customer is not. Make its work visible, with a log of what it did and why, and a regular review. And give it an owner - a named person accountable for what the agent does, exactly as for a new hire. The businesses that get the most from always-on agents will be those that treat them like capable new colleagues: given clear objectives, appropriate access and proper supervision.

What the UK exclusion actually signals: It would be easy to read the UK's exclusion from the Dots Pro rollout as Britain being left behind. It is better read as a signal about where the regulatory questions lie. An agent that acts autonomously across thousands of apps, on personal and business data, raises exactly the issues UK GDPR and the Data (Use and Access) Act care about: what data it processes, on what lawful basis, with what transparency, and how people can challenge automated decisions. A vendor shipping first where those questions are less pressing is a commercial choice, not a verdict on the technology. The practical point for UK businesses is that they will need to answer those questions themselves for any always-on agent they deploy - which is an argument for building with governance designed in from the start, rather than waiting for a consumer product to arrive.

How A UK Business Can Build This Today

The good news is that the always-on pattern does not belong to one vendor or one product. It is an architecture - an agent with a goal, persistent memory, scheduled and event-triggered runs, connections to your tools, and human checkpoints for the decisions that matter - and it can be built today on platforms a UK business controls, with data kept where it should be and governance designed in. The practical path is to pick one piece of recurring work that eats real hours, define the objective and the boundaries clearly, connect the agent only to the systems it needs, set it running on a schedule or on the events that should trigger it, keep a person in the loop for consequential actions, and review its log weekly. Prove it on one job, then extend. Done this way, a UK business gets the leverage of always-on agents now, on its own terms, with compliance built in - rather than waiting for a product that may arrive later and on someone else's terms. That is precisely the work we do as an AI Agency London.

The Bottom Line

OpenAI's Dots are the clearest sign yet that AI is moving from a tool you visit to a colleague you delegate to: always-on agents powered by GPT-6 Astra, connected to more than 4,000 apps, working through Slack and Teams, taking goals instead of prompts and learning over time. Their value lies in recurring, multi-step, cross-tool work that needs persistence rather than genius - reporting, pipeline upkeep, monitoring, triage, coordination - and their risk lies in the same breadth and autonomy, which demands scoped access, approval gates for consequential actions, visible logs and a named owner. The UK's exclusion from the Pro rollout is a regulatory signal, not a technological one, and it means British businesses must answer the data and accountability questions themselves. The opportunity is that the always-on pattern can be built now, on platforms you control, with governance designed in. The businesses that start delegating recurring work to well-governed agents this year will have a capacity advantage that compounds - and helping UK businesses build exactly that is what we do.

References & Further Reading

Emergent - OpenAI DevDay 2026: every announcement, from Dots to GPT-6.1 Sol (availability and UK/EEA exclusion): https://emergent.sh/news/openai-devday-2026
Latent Space - AINews: OpenAI DevDay 2026 - Dots, 6.1 Sol, Ultrafast, Decisions API, Agents API, Spaces, Marketplace: https://www.latent.space/p/ainews-openai-devday-2026-dots-61
The Neuron - OpenAI DevDay 2026: Dots and ChatGPT's agent OS: https://www.theneurondaily.com/p/openai-launched-dots-20-more-tools
Pulse 2.0 - OpenAI unveils Dots, GPT-6.1 Sol, $500 Pro tier and new enterprise AI tools: https://pulse2.com/openai-unveils-dots-gpt-6-1-sol-500-pro-tier-and-new-enterprise-ai-tools/
AI Weekly - AI news for October 2, 2026: https://aiweekly.co/ai-news-today/edition/2026-10-02