Time awareness and async input for personal agents
Blog post from OpenAI
The text explains how to build a meeting-preparation agent that researches a company, asks for useful context, and still delivers a briefing on time if the user does not respond. In its example, the agent prepares for a meeting in 15 minutes and must provide a briefing 12 minutes from the start, leaving time for review; it asks whether the meeting is technical or commercial while continuing independent research. The Agents API’s time-awareness feature helps the model adjust its research and writing priorities as the deadline approaches, while an asynchronous input tool lets the application collect a reply without blocking the agent’s work. The application displays the question, submits the user’s answer—or a no-answer result after a cutoff—and retrieves and delivers the completed briefing, or provides the latest available findings if the deadline arrives first. The text also describes implementing the same pattern in a hand-built Responses API loop, where each request includes an updated timestamp and an application timer enforces delivery, with pending input and stored responses cleaned up afterward. Examples are provided across several programming languages and SDKs.
No tracked trend matches for this post yet.
Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.