Run shopping, trip planning and project tracking with a dot
Pat Simmons gave OpenAI's new Dots agent ordinary work: shop for a few things, plan a trip, track a training project, monitor his coding agents, and help with business research. Task: Pat Simmons assigned OpenAI's Dots agent tasks including shopping, travel planning, training tracking, and business research. The agent acted as a persistent coordinator, generating travel research, a training dashboard, and an unsolicited billing alert. Human involvement was critical for approving actions, as the agent reached a cart but did not complete purchases, and required manual intervention to bypass browser blocks.; tools: Dots, ChatGPT Agent. Human role: The original post does not say how a person stayed involved.; limitations: Dots is limited to eligible Pro users outside the EEA, Switzerland, and UK. It cannot initiate calls or bypass site access restrictions without explicit local browser permissions, which expose security risks. Disconnecting apps does not delete previously obtained data. The agent's memory degradation over time and accuracy of future scheduled tasks remain untested..
Tools
Published on Agent Work Atlas:
Overview
Pat Simmons gave OpenAI's new Dots agent ordinary work: shop for a few things, plan a trip, track a training project, monitor his coding agents, and help with business research.
Workflow
Pat Simmons assigned OpenAI's Dots agent tasks including shopping, travel planning, training tracking, and business research. The agent acted as a persistent coordinator, generating travel research, a training dashboard, and an unsolicited billing alert. Human involvement was critical for approving actions, as the agent reached a cart but did not complete purchases, and required manual intervention to bypass browser blocks.
Reported result
The agent produced useful research and status updates but failed to complete transactions or bookings. It generated a rough training dashboard and identified project issues, but outputs like buyer lists and newsletters were unverified or generic. No long-term reliability or completed sales were demonstrated, highlighting a gap between output volume and actionable results.
Human role
The original post does not say how a person stayed involved.
Limitations
Dots is limited to eligible Pro users outside the EEA, Switzerland, and UK. It cannot initiate calls or bypass site access restrictions without explicit local browser permissions, which expose security risks. Disconnecting apps does not delete previously obtained data. The agent's memory degradation over time and accuracy of future scheduled tasks remain untested.
Try it
Act as a workflow tester for OpenAI's Dots agent. I will provide [replaceable inputs: specific business context, target user, and source files/URLs]. Your goal is to execute a bounded, read-only monitoring task. First, perform a dry run on a small sample dataset to verify data access and output format. Do not execute any purchases, bookings, or message sends. Generate a report summarizing findings, including any failed alerts or blocked actions. Wait for my explicit confirmation before attempting any active tasks. If inputs are missing, ask clarifying questions. Measure success by the accuracy of the dry-run report and the clarity of approval requests. Ensure all outputs are labeled as 'Suggested implementation' if no execution evidence exists.
Source reports are not independently verified. Publication date is the catalog publication date, not the original article date.