Particle.news

OpenAI Lets ChatGPT Run Multi-Step Workflows by Voice

The update aims to make spoken commands a practical way to complete documents, emails and cross-app tasks with a single assistant.

Overview

  • OpenAI rolled out voice-based agent features on mobile and web on Wednesday, linking its GPT-Live conversational audio to agentic workflows that can trigger multi-step tasks in browsers and apps.
  • Users can now ask ChatGPT Voice to draft documents, summarize emails or Slack threads, build spreadsheets and run background browser research without typing.
  • Plus and Pro subscribers get the Work tab on mobile for voice-driven creation while Free and Go users gain access to plugins and connected apps like Gmail, Calendar and Slack.
  • Some outlets reported the update uses OpenAI’s new GPT-6 model suite, but that specific technical claim has not been confirmed across the wider coverage and remains unverified.
  • OpenAI added admin controls for Enterprise and education accounts to require review of sensitive actions, a design choice that responds to privacy and security concerns as the assistant gains access to inboxes, calendars and financial tools.