OpenAI Introduces Dots. Now businesses need to put them to work.
OpenAI’s new Dots are always-on AI agents designed to work continuously, connect to business applications and take responsibility for real work. The technology is arriving. The next challenge is helping organisations understand where Dots belong, how to configure them and how their people should work with them.
OpenAI Dots in brief.
On September 29, 2026, OpenAI introduced Dots — a new generation of always-on AI agents designed to continuously work on behalf of people and organisations.
A Dot is not simply another chatbot waiting for someone to open a window and type a prompt.
According to OpenAI, Dots have their own cloud computer, can work towards goals 24/7, learn from feedback over time and connect through an ecosystem of more than 4,000 applications.
They can interact with users through ChatGPT, Slack and Microsoft Teams, monitor ongoing work, investigate issues, prepare outputs and ask for input or approval when necessary.
OpenAI is also beginning to test specialist Dots designed to take on dedicated responsibilities inside organisations.
This is an important change.
AI is moving from something employees occasionally use to something businesses may increasingly delegate ongoing responsibilities to.
From AI Assistant to AI Co-Worker.
For most organisations, the first generation of generative AI has been based around individual tasks.
Write an email. Summarise a document. Analyse a spreadsheet. Research a company. Prepare a presentation. Generate a report.
These are useful productivity improvements.
But the employee still remains responsible for starting the work, moving between systems, coordinating the next step and deciding when something needs to happen.
Dots introduce a different model.
Instead of asking AI to complete isolated tasks, an organisation can begin thinking about giving AI a defined objective and allowing it to keep working towards that objective.
A Dot could monitor something. Investigate something. Follow a project. Prepare recurring work. Work across connected applications. Ask questions when required. And return when human attention is needed.
AI is moving from something people use to something organisations can increasingly delegate work to.
Almost every business has work a Dot could help with.
The opportunity is not limited to software companies or highly technical teams.
Most organisations contain work that depends on people repeatedly checking systems, gathering information, coordinating activity, preparing recurring outputs and following up actions.
These are exactly the areas where agentic AI becomes interesting.
The question for businesses is therefore not whether Dots are interesting.
The important question is where they can create measurable operational value.
Access to AI is not the same as AI adoption.
This is where many organisations risk repeating the same mistake made during the first wave of generative AI.
A new technology becomes available.
Management buys licences.
Employees receive access.
Some experiment extensively. Others use it occasionally. Different departments develop different approaches. And very little is measured.
Eventually the organisation can say that it “uses AI”.
But having access to AI and transforming the way a company works with AI are two very different things.
Buying the technology is the easy part. Changing how the business works is where the real transformation begins.
Someone still has to design how the Dot works.
OpenAI can provide extremely capable technology.
But no technology platform automatically understands how your organisation should operate.
Before deploying Dots inside a business, someone needs to answer much more practical questions.
What should the Dot actually be responsible for?
Which department should use it first?
Which processes should it monitor?
Which applications should it connect to?
What information should it be allowed to access?
Which actions should it perform independently?
Which actions should require approval?
Who reviews the work?
What happens when something goes wrong?
How should employees interact with it?
And most importantly: how will the organisation measure whether it is actually creating value?
Process discovery, responsibilities, integrations, permissions, approvals, employee training and measurable business outcomes all need to be designed around the technology.
AI training is changing too.
Teaching employees how to use AI can no longer mean simply showing them a list of prompts.
As AI systems become more agentic, employees will need to understand how to delegate work, provide context, define objectives, establish boundaries and review the work produced by AI.
Managers will need another level of understanding.
They will need to know which responsibilities can be redesigned, where human approval remains necessary, how agent activity should be monitored and how AI changes the roles of the people inside their teams.
The question is moving from:
“How do I use ChatGPT?” is becoming “How do I manage work when AI agents are part of my team?”
That is a very different organisational capability.
The competitive advantage will not be the Dot.
If Dots become broadly available, thousands of businesses will eventually have access to the same underlying technology.
That means access itself cannot become a sustainable competitive advantage.
Imagine two companies using exactly the same AI platform.
The first company gives employees access and allows everyone to experiment individually.
The second identifies ten high-friction workflows, redesigns them, connects the necessary systems, defines responsibilities and permissions, trains its employees and continuously measures the results.
Same underlying technology.
Completely different business outcome.
The Dot itself is not the competitive advantage. The advantage is knowing what to do with it.
From “We have AI” to “AI works here.”
The real transformation starts when AI becomes part of the operating model of the organisation.
When the company can clearly say:
Our finance Dot handles these responsibilities.
Our sales Dot monitors these processes.
Our operations Dot watches these systems.
Our management Dot prepares this intelligence.
Our people understand how to work with them.
Our permissions and approval processes are defined.
And we can measure the impact.
At that point, AI is no longer simply a tool or an experiment.
It has become part of how the organisation operates.
OpenAI builds the technology. Shift Works helps businesses put it to work.
At Shift Works, we do not believe businesses need more AI simply because new AI technology exists.
They need to understand where it creates value.
A company does not simply need someone to activate a Dot.
It needs someone who can understand how the organisation works, identify where Dots should be used, define what they should be responsible for, connect the right systems, structure permissions and approvals, train employees and managers, redesign the surrounding processes and measure the business outcome.
That requires business transformation, process redesign, AI training, integration and implementation working together.
That is exactly the gap Shift Works was created to address.
Do not start with “How do we get a Dot?”
Start with a more important question:
If AI could continuously work alongside our people, where would it create the most value?
Understand the work.
Identify the opportunity.
Redesign the process.
Prepare the people.
Configure the technology.
Measure the outcome.
Then scale what works.
The future of enterprise AI will not belong to organisations that simply have access to the most advanced agents.
It will belong to organisations that know what to do with them.
OpenAI’s official announcement introducing always-on Dots and specialist Dots for organisations, published September 29, 2026.