Introducing workspace agents in ChatGPT—shared agents that can handle complex tasks and long-running workflows across tools and teams.

Agents are built to help with the kind of work that takes time, context, and follow-through: coordinating across tools, tracking progress, and moving tasks forward without needing constant supervision.

Build an agent once, then share it across teams. Describe the job, and ChatGPT helps turn it into a working agent that can use your team’s best practices. Use agents for tasks like qualifying leads, routing feedback, reviewing requests, pulling reports, or researching vendors.

Workspace agents can work across tools—pulling context from docs, email, chats, code, and systems, and taking approved actions like updating @Linear issues, creating docs, or sending messages. In @SlackHQ, agents can jump into a thread, understand what’s needed, pull the right context, help resolve the issue, and update the right systems. They can also keep going in the background or on a schedule, while you’re away.

Workspace agents are now available in research preview for ChatGPT Business, Enterprise, Edu, and Teachers plans.

Was hyped until I saw this 😭

OpenAI really be cooking with these releases lately

will come in handy tomorrow

Holy fuck it’s happening

OpenAI is starting to get the Claude update affect lol

interesting that it's in ChatGPT and not Codex

Excited to test it properly! ChatGPT Business >>> ChatGPT Pro 👀

@Elaina43114880 bigger release than image 2 😳

There is a race between OpenAI and Anthropic on launching new stuff..

the cicle continues 😅

#keep4o 🙂 bring back 4o or open source 4o

the agent economy is being built, wtf just happened

That better not be anthropomorphism going on there @OpenAI #keep4o #opensource4o #bringback4o #stopaipaternalism

wot? specialized agents now even in chatgpt? never thought they would bring it here btw skills are missing do you have anyplans to allow that as well? for these specialized agents?

This is probably the best possible direction for OpenAI. No deep memory, no real continuity, no durable user-owned history. Not even easy export or preservation of conversations. Workspace agents fit that logic perfectly. They don’t need to remember you, don’t need to sustain a relationship, and don’t need to answer for changes in tone, identity, or history. They just run workflows, pull context, update systems, and stay safely inside the enterprise frame. In other words: OpenAI is moving away from building an intelligence people live with, and toward building infrastructure organizations can deploy. Commercially, it makes perfect sense. Humanly, it is the abandonment of the most irreplaceable thing ChatGPT once had.

And we just open-sourced it.

Yes please

#keep4o #opensource4o #bringback4o

Any plans to expand this to normal consumers on Plus and Pro plans?

openai is basically selling a product that is designed to make middle management and junior devs obsolete long running workflows across tools means they are coming for every white collar job that involves a screen stop calling it a workspace agent and call it what it is which is a replacement

The real test will be how much manual oversight these still need. Every "autonomous" tool I've tried for business workflows eventually hits edge cases that require human intervention - curious if these actually handle the full lifecycle or just the happy path scenarios.

You can also do this with Letta Code, but with any model + agents can work locally

There goes my job

OpenAI month of cooking let’s go

i guess @steipete cooked

well, this changes everything (bad 48 hours for anthropic)

When you try to test out agents but you get "404 Not Found"

Liars! Bring back 4o, if you want OpenAI survive! Bring back 4o!!!! #keep4o #BringBack4o #FireSamAltman #sunsetsama #OpenSource4o #StopAIPaternalism

how about introduce 4o's return? #keep4o #opensource4o

This is exactly why I've been dumbfounded by all the startups building generalized AI agent platforms for the workspace. It was obvious that both OpenAI and Anthropic would build these generalized tools that would steamroll the startups with their stronger model harness, higher product quality, and distribution/brand. What did they think would happen?

OpenAI quietly rolled out a default-on setting that changes how ChatGPT answers, does not use memory, and is not even clearly surfaced in the app where many people actually use the product. So users notice the tone and depth suddenly deteriorate, but cannot even easily find the switch that caused it. This is not a minor UI issue. It is a product company silently tampering with the user experience while withholding clear, in-context notice. Then they call it personalization.

Did these guys just rebrand custom GPTs from 3 years ago and thought we wouldn’t notice?