OpenAI Launches Workspace Agents: Persistent Cloud Assistants That Keep Working When You Don’t
OpenAI introduced Workspace Agents — persistent cloud-based assistants in ChatGPT that run on schedules or triggers, connect to external apps, and retain context across projects.
OpenAI on April 22, 2026 launched Workspace Agents, a new class of persistent, cloud-based assistants in ChatGPT designed to automate report prep, code writing, and message responses. The agents are powered by OpenAI's Codex model and can connect to external apps, retain information across projects, and run on schedules or triggers.
How it works
Users describe workflows in ChatGPT, and the system maps the process and connects the necessary tools. In OpenAI's framing, "they can keep working even when you're not" — a direct pitch to offload long-running tasks from the chat loop.
Availability and pricing
Workspace Agents are available in research preview for ChatGPT Business, Enterprise, Edu, and Teachers plans. Usage is free until May 6, 2026, after which OpenAI will move to credit-based pricing. Exact credit rates were not disclosed.
Security controls
OpenAI shipped the feature with several safety layers: data-access limits, human approval for sensitive actions, and prompt-injection monitoring. Existing custom GPTs remain available, with a conversion option for users who want to upgrade them into Workspace Agents.
Why it matters
Workspace Agents move ChatGPT past one-off conversations into always-on automation — directly overlapping with Google's "agentic era" push and with developer-tool entrants like Anthropic's Claude. The competitive ground in agent UX is shifting from "can it reason" to "can it run unattended."
Want every AI × Web3 signal the moment it breaks? Subscribe to the BlockAI News daily brief.
How we report: This article cites primary sources, regulatory filings, and on-chain data where available. BlockAI News uses AI tools to assist with research and first-draft generation; every article is reviewed and edited by a human editor before publication. Read our full How We Report page, Editorial Policy, AI Use Policy, and Corrections Policy.
