OpenAI’s new Dots agents could change the economics of office work by turning AI from a tool employees use into software that performs workflows itself.
The most important question surrounding AI is changing.
It used to be:
How smart is the model?
Now it is increasingly:
How much work can the model actually complete?
OpenAI’s Dots are designed around that second question.
Why It Matters
Most businesses already use dozens of software applications.
But applications don’t do the work by themselves.
Employees do.
A salesperson operates the CRM.
A marketer operates analytics software.
A manager operates Slack.
An accountant operates spreadsheets.
A recruiter operates an applicant-tracking system.
AI agents could change that.
Instead of employees operating every application manually, an agent could potentially operate the applications on their behalf.
The Digital Workforce
Imagine an employee receiving an instruction:
“Prepare this week’s customer-retention report.”
Today that might involve:
- Exporting CRM data
- Cleaning spreadsheets
- Checking support tickets
- Reviewing emails
- Creating charts
- Writing a summary
- Sending it to management
An AI agent could potentially coordinate many of those steps.
That doesn’t necessarily eliminate the employee.
It changes what the employee does.
The human becomes responsible for:
Goals → judgment → approval → exceptions.
The AI handles more of the execution.
Productivity Could Become the New AI Metric
AI companies have spent years advertising benchmark scores.
But businesses don’t pay for benchmarks.
They pay for outcomes.
If an AI agent saves an employee five hours per week, that has measurable economic value.
If it saves 20 hours, the value becomes much larger.
That could make agentic AI easier for businesses to justify financially than generic chatbot subscriptions.
Why Software Companies Are Watching
This is potentially disruptive to enterprise software.
If employees increasingly access software through AI agents, traditional applications could lose some direct user interaction.
The agent becomes the interface.
That could affect:
- Salesforce
- ServiceNow
- Microsoft
- Atlassian
- Adobe
- CRM platforms
- HR software
- Accounting software
The question isn’t whether those companies disappear.
It is whether they remain the primary interface.
The New AI Stack
The emerging architecture could look like:
AI model → agent → tools → applications → data
Instead of:
Human → application
That creates a new layer between people and software.
OpenAI is building this layer.
Meta is building it.
Google is building it.
Anthropic is building it.
The race is therefore becoming much bigger than chatbots.
The Computing Cost
There is another side to the story.
An AI that works continuously consumes more compute than one that responds to occasional prompts.
Persistent agents can create additional demand for:
- GPUs
- CPUs
- Memory
- Storage
- Networking
- Cloud infrastructure
That means the agent economy could become another major driver of data-center investment.
What Happens to Jobs?
This is where the debate becomes complicated.
AI agents can automate tasks.
But jobs consist of collections of tasks.
A company may therefore automate part of a job without eliminating the entire role.
The more likely near-term transformation could be:
fewer repetitive tasks + more human oversight + higher expectations for output.
The impact will vary substantially by occupation.
The New Competitive Advantage
Businesses may eventually compete on how effectively they deploy AI agents.
Two companies could have similar employees and similar software.
But one could automate 70% of its repetitive workflows while the other automates 10%.
That difference could affect operating costs and speed.
AI agents could therefore become a productivity infrastructure rather than simply an employee benefit.
What’s Next
OpenAI’s Dots are still an early implementation of this idea.
But the direction is clear.
AI is moving from:
Answering questions
to:
Completing tasks
to:
Managing workflows
to potentially:
Running entire business processes.
The company that wins the agent race may not necessarily have the chatbot with the highest benchmark score.
It may have the AI that businesses trust to complete the most valuable work.
FAQ
What are OpenAI Dots?
Dots are always-on AI agents designed to complete multi-step tasks and work with connected applications.
Could Dots replace employees?
They could automate portions of some workflows, but the impact on entire jobs will vary by occupation and business process.
Why are AI agents important for software companies?
Agents could become a new interface layer between users and traditional software.
Will agents increase AI infrastructure demand?
Potentially. Always-on agents require computing resources to monitor information, reason, use tools and execute tasks.
Related Reading
- OpenAI Dots: Why ChatGPT Is Becoming an AI Agent
- Meta Muse: The Personal AI Agent Race
- Why AI Infrastructure Spending Is Exploding