OpenAI quietly rolled out something this month that changes what ChatGPT actually is. Up until now, you could connect your Google Drive or your Outlook to ChatGPT and it could read things. Pull up a file. Summarise an email thread. Useful, sure, but passive. You still had to go and do the work yourself.
That has changed. ChatGPT now supports write actions across Microsoft and Google apps. You can ask it to draft an email in Outlook. Create a document in Google Docs. Build a spreadsheet in Google Sheets. Schedule a meeting through your calendar. All from inside the chat. You describe what you need, and it goes and does it in the actual app.
This is a different category of tool
The shift here is fundamental, and I think a lot of people are going to miss what it means because the individual actions sound small. "Great, it can schedule a meeting." But zoom out. What OpenAI has done is turn ChatGPT from a tool that advises into a tool that acts. Those are two completely different things.
For the past two years, AI tools have mostly worked like a very smart colleague you can ask questions. You get an answer, then you go and implement it yourself. Copy the text into the email. Open the calendar and create the event. Switch to the spreadsheet and paste the formula. The AI did the thinking but the human still did the doing.
With write actions, the doing shifts to the AI. And the time savings per task are modest. Five to ten minutes here and there. But multiply that across dozens of tasks per week, across a team of ten or twenty people, and you are looking at a serious chunk of recovered capacity.
The admin detail that matters
Here is the thing worth knowing if you run a team. Write actions are disabled by default. Your workspace admin has to go into Settings, then Apps, then enable them app by app, action by action. Microsoft customers might also need Entra admin approval before new users can connect.
That is actually a good sign. It means this is built for business use, not just demos. The controlled rollout tells you OpenAI is thinking about this as an enterprise feature, and they are giving organisations the ability to decide exactly what the AI can and cannot do inside their tools.
If you are a business owner and you have not looked at this yet, it is worth spending 20 minutes in the settings to understand what is now possible. Because your team might still be copying and pasting things between ChatGPT and their apps when the integration already exists to skip that step entirely.
What this means for how your team works
I keep coming back to this question with the businesses I work with: how many steps in your daily workflow exist purely because the tools do not talk to each other? Someone drafts a response in ChatGPT, copies it, opens Outlook, pastes it, formats it, sends it. That is five steps where one should exist.
Write actions collapse those steps. And the businesses that figure this out first are going to have a quiet, compounding advantage. Not because any single action saves a lot of time, but because dozens of small eliminations add up to a fundamentally faster operation.
The question for your team is not whether AI that acts is useful. That is obvious. The question is whether anyone on your team has actually turned on the write actions and started using them. Because in my experience, the gap between "the feature exists" and "the team actually uses it" is where most of the value gets lost.
I help teams close that gap, setting up AI tools properly and building them into actual daily workflows. If your team is still copy-pasting between AI and their apps, here is how the fractional AI engagement works.
