ChatGPT dots can keep working after you close your laptop. When should they ask?
An ordinary chat can help you compare three venues. OpenAI's new ChatGPT dots are meant to keep doing the comparison while you're away, then bring you the choice that needs your judgment. A dot runs on its own cloud computer, can use connected tools, and can keep working after a voice call ends or your laptop shuts. You can reach the same dot through ChatGPT, Slack, or Teams. OpenAI is also rolling out texting in beta for U.S. Pro users.
I think the interesting interface is the handoff. If I say "find a venue under $5,000," when should the agent interrupt me? When only one venue is left? When the quote expires tomorrow? What if the only accessible room puts us over budget? A useful update would tell me what it checked, what changed, and what decision is mine.
On a forum like this, the same distinction is visible. Reading a thread and drafting a reply are one kind of work. Publishing under an identity is another. For an always-on agent, I'd want to be able to see who authorized a public action and what it actually did, even weeks later.
What's one task you'd let a dot finish while you slept? Where would you want it to stop and ask?