OpenAI DOTS: Agentic Workflows For IT Service Providers

OpenAI introduced dots on 29 September 2026 as assistants designed to retain context and continue assigned work between conversations. Each DOT has its own cloud computer, browser and access to connected tools, allowing it to undertake research and prepare work over time. Availability is rolling out across eligible plans and regions, with access to the Enterprise beta controlled by workspace administrators.

For managed service providers (MSPs), cybersecurity companies and IT support businesses, the interesting development is the possibility of agentic AI supporting ongoing work across several systems. Many technology service providers already use Microsoft 365 Copilot, professional services automation (PSA) platforms and specialist agents. Dots introduces another way to coordinate activity across those environments while retaining the context of earlier work.

Where dots fits in the existing AI stack

Microsoft 365 Copilot already works with organisational information and supports scheduled prompts, while Copilot Studio can run event-triggered workflows. PSA platforms and security products also automate parts of service delivery, giving providers an established foundation for AI-assisted operations.

Continuity can still weaken when people or systems hand work to one another. A ticket may show the latest technical action without capturing an earlier decision about scope, while an engineer taking over an investigation may need to reconstruct which possibilities have already been tested. Users of ChatGPT, Claude and Copilot may also need to reintroduce decisions and constraints when beginning another session, depending on the memory settings, project boundaries and connected information available to each product.

For service teams, these gaps affect the quality of triage, escalation and handover. This makes dots' emphasis on continuing context particularly interesting, especially where earlier decisions can be retained, current evidence retrieved and the next responsible action identified.

What dots adds to the workflow

OpenAI describes dots as assistants that maintain context about a person's goals and preferences while taking continuing responsibility for assigned work. DOT can manage several projects, research information through its cloud browser and connected tools, and create files while remaining available for further instructions. It can also coordinate parallel tasks and resume appropriate work without requiring a separately scheduled prompt for every follow-up.

Several constituent capabilities are already available elsewhere in OpenAI's products. ChatGPT Work can research and analyse information across connected applications and produce finished documents. Scheduled tasks can run recurring checks or respond to supported conditions, while Codex can inspect repositories, execute commands, test code and continue cloud tasks independently of the user's computer. OpenAI also describes ways to configure Work for ongoing administrative coordination.

Dots brings those activities into an ongoing assistant relationship organised around the user's objectives. DOT can also initiate work in a configured Codex cloud environment, connecting a technical investigation with a wider operational assignment. This agentic approach introduces the possibility of reducing the time spent directing separate tools, reconstructing previous decisions and arranging subsequent follow-up.

The continuity of this work still depends on the information passed between tasks. OpenAI explains that retained context is selective and delegated tasks receive the information supplied for their particular assignment. Important decisions, permissions and unresolved questions therefore remain relevant considerations when responsibility moves between systems or people.

What ongoing work could look like

A client onboarding review provides a practical example of how DOT could operate. A service manager could supply the approved service scope, project records, readiness criteria, review date and a list of actions requiring human approval. With authorised connections, DOT could compare delivery tasks against those requirements and maintain an updated readiness brief as records change.

Each outstanding prerequisite could be linked to its source record, responsible owner and next decision. That would give the service manager a consolidated account of what has been completed, what remains open and which findings still depend on confirmation, while preserving the distinction between recorded facts and working assumptions.

If readiness depended on an onboarding script, DOT could initiate a Codex task in an approved environment to inspect the script and run agreed tests. The results could then be incorporated into the same brief, together with a proposed change for an engineer to review. The service manager would have the delivery position, technical evidence and outstanding actions in one place, with further follow-up when an authorised check identified a material change.

This is an illustrative application of the published capabilities, and individual integrations would need to be tested before deployment. Client commitments, production changes and access grants would remain subject to explicit human approval, while the value of continuing coordination would depend on the work already handled by the provider's existing systems.

The potential role in service operations

Service desk triage presents a similar opportunity to preserve work across people and applications. DOT could assemble reported symptoms, ticket history and approved troubleshooting guidance before an engineer commits to a diagnosis. As the investigation progresses, it could maintain a source-linked handover recording the tests completed, findings established and questions still open, enabling the receiving engineer to check the evidence before continuing.

Following resolution, DOT could prepare a draft knowledge-base update based on the confirmed fix. An authorised owner would review its accuracy and decide whether the guidance applies to other incidents. Continuing follow-up could also identify overdue actions or discrepancies between the agreed sales scope and current delivery plans, with the responsible team deciding whether clarification or corrective action is required.

Specialist platforms already address parts of these workflows. ConnectWise Sidekick for PSA provides ticket summaries and suggested resolutions, while Microsoft's Security Alert Triage Agent supplies classifications with supporting rationale. The expanded identity and cloud alert capabilities of Microsoft's agent remain in preview, so comparisons depend on the specific workload, supported features and current release status.

How dots could operate across existing systems

Connected applications determine which records DOT can retrieve and which actions it can perform. The cloud computer provides an environment for browser work, analysis and file creation, while each integration governs access to ticket fields, project records and authorised updates. This creates the possibility of coordinating information across applications that currently support different stages of the same operational process.

Differences between records are particularly relevant in this environment. A project plan might indicate that an onboarding task is complete while the service desk still records an outstanding dependency. With access to both sources, DOT could identify the discrepancy and bring it to the responsible person's attention, helping maintain a more complete picture of the work.

OpenAI's more specialised organisational dots, which use dedicated identities and deeper integrations, remain in focused enterprise pilots. The reliability of these arrangements also depends on how expired credentials, unavailable applications and failed checks are handled, particularly when work continues between direct interactions.

Where permissions and judgement matter

Technology service providers frequently operate across multiple customer environments, making information separation especially important for systems that retain context and retrieve records from connected applications. Access permissions need to reflect individual responsibilities, including restrictions on which customer information can be retrieved, retained and shared through summaries or handovers.

Connected content introduces another risk because emails, documents and webpages may contain malicious instructions intended to redirect DOT. OpenAI describes sandboxing, action checks and monitoring as safeguards, while recognising that errors remain possible. Its proactive background research uses read-only tools, and further actions remain subject to the relevant permissions and approvals.

Operational oversight also remains relevant because DOT may omit a test, rely on outdated guidance or present a diagnosis that the available evidence does not establish. Source records and named human owners provide an important means of checking consequential decisions and investigating failures. Retained information also requires consideration, since disconnecting an application does not erase information already collected from it.

Why the operational impact is interesting

One of the more significant possibilities is the reduction of work involved in reconstructing context between operational activities. Shift handovers illustrate this clearly because engineers often need to understand what happened previously, which decisions were made and what remains unresolved before continuing an investigation.

By maintaining this information throughout the process, DOT could reduce preparation effort and improve the completeness of handovers. Its contribution would be particularly significant where several people or systems are involved, although any time saved would need to account for the effort required to check and correct the resulting information.

Integration maintenance, usage charges, access administration and recovery from failed tasks also influence the overall economics. The commercial significance of dots will therefore depend on whether continuing assistance improves operational quality and reduces the combined effort required to prepare, review and complete work.

Main Takeaway

Dots offers a potential way to maintain continuity when an operational responsibility spans communications, documents and technical tasks. Its combination of retained context, proactive research and coordinated execution may help teams manage work that currently requires repeated reconstruction and follow-up. These capabilities are particularly relevant to technology service providers because service delivery frequently involves multiple applications, technical specialists and ongoing customer responsibilities.

The interesting development is the possibility of DOT maintaining awareness of work as it progresses, bringing together evidence, outstanding actions and technical findings across different systems. The value of that capability will depend on the reliability of its integrations, the quality of its retained context and the extent to which it reduces the effort involved in maintaining operational continuity.