All content
Assistants
15 Mar 2026

ChatGPT can now act inside Outlook, Google Docs, and your calendar.

AI is not getting more capable slowly. It is accelerating. ChatGPT Business can now draft your emails, create documents, and book meetings. Write actions mark a shift from AI that advises to AI that acts. The businesses that benefit most will not be the ones resisting this. They will be the ones building the fluency to guide, edit, and review AI well. Here is what changed, why the default-off design is the right call, and three steps admins should take this week.

Written by
The gecco team

We can all see AI becoming more capable around us. It is not a question of if, but when AI will outperform people in more and more domains. That is not a comfortable thing to say. But pretending otherwise does not change the trajectory.

The right response is not resistance. It is fluency.

The people who will be least affected by this change are the ones who accept it now, learn how to guide AI well, and position themselves as the ones directing the work rather than competing with it. That does not require technical expertise. It requires honesty about what is changing and a decision to get ahead of it.

Which brings us to this month's ChatGPT Business update.

For the past two years, AI assistants have been excellent at telling you what to do. Now they are starting to do it.

OpenAI's latest ChatGPT Business release notes confirm what many businesses have been waiting for: write actions. Connected apps like Microsoft Outlook, Google Docs, Google Sheets, and calendar tools can now draft emails, create documents, and book meetings, all from inside a ChatGPT conversation.

That is a meaningful shift. And it is worth understanding both what it means and what it does not mean.

What write actions actually are

Until now, connectors in ChatGPT Business worked in one direction. ChatGPT could read your Google Drive, search your SharePoint, or pull context from Slack. It could find the answer and hand it back to you.

Write actions change the direction. ChatGPT can now act on your behalf, inside the tools you already use, based on what you ask it to do.

Ask it to draft a follow-up email from your last meeting notes. Ask it to create a project tracker in Sheets from a bullet list. Ask it to schedule a call based on your calendar availability. These are tasks that previously required you to leave the conversation, open another tool, and do it yourself.

The time saved per task is small. Ten minutes here, five minutes there. Across a team of 20 people doing this dozens of times a week, it adds up fast.

The trust mechanism that makes this work

Here is the part of the release notes that most people will skip over. Write actions are disabled by default.

Workspace admins must deliberately enable them, app by app, action by action, through Workspace settings. For Microsoft tools, IT or Entra admins also need to review and approve updated permissions before users can connect.

That friction is intentional. And it is the right call.

The biggest barrier to AI adoption in UK SMEs is not cost or complexity. It is trust. Business owners and team members want to know that AI is not going to send an email they did not approve, overwrite a document by mistake, or create a calendar invite at the wrong time with the wrong people.

The default-off design answers that concern directly. You decide which actions to unlock. You review what each app is permitted to do. You control the pace.

This is how AI earns the right to do more over time. Not by asking for all permissions upfront, but by proving itself in contained, reviewed steps.

What this looks like in practice

A useful way to think about this is in two phases.

Phase one is read-only. ChatGPT searches your tools, surfaces information, and helps you decide what to do. This is where most businesses using ChatGPT Business currently sit.

Phase two is read-and-act. ChatGPT does the donkey work inside your tools, with your approval, while you focus on the decisions that genuinely need your attention.

The March 2026 update moves ChatGPT Business firmly into phase two territory. But it does so without forcing you to jump straight in. You can enable Outlook email drafting today, leave Google Sheets write access off for now, and add more when your team is ready.

That graduated approach matters. It mirrors how trust actually builds in any organisation.

What admins should do now

If your business is on ChatGPT Business, three practical steps are worth taking this week.

First, review your current connector setup. Visit Workspace settings, open the Apps section, and check which apps are connected and what actions are currently enabled. Many admins have not looked at this since connectors were first switched on.

Second, identify one or two low-risk write actions to pilot. Email drafting is a good starting point. The output is always reviewed before sending, which means the risk of an error reaching a client is low. Confidence builds quickly from there.

Third, brief your team before you enable anything. People respond better to AI tools when they understand what the tool is doing and why. A five-minute explanation at a team meeting reduces the hesitancy that slows adoption.

The broader direction of travel

This update is part of a longer arc. Over the past year, ChatGPT Business has added project sharing, company knowledge, synced connectors across 20-plus platforms, and now write actions. Each update moves the product further from a smart search tool and closer to an active participant in day-to-day work.

That is the direction AI is heading for every business, regardless of which platform you use. The question is not whether to engage with it. The question is whether you are building the internal confidence and governance to use it well when it arrives.

Businesses that have already invested in AI training, set clear usage policies, and developed internal AI champions are in a much stronger position to adopt features like write actions safely and quickly. Businesses that are still at the "let's see how it goes" stage will find the pace of change increasingly difficult to manage.

The tools are moving faster than most teams are ready for. Closing that gap is where the real work sits.

The best people are not waiting to see how this plays out. They are learning now, so they are capable of affecting the change itself rather than simply being affected by it.

Adopt AI Today
Insights
17 Mar 2026

Claude's Cowork just got untethered from your desk

Anthropic's Cowork now lets you assign AI tasks from your phone while Claude works on your desktop. For SMEs already using Claude, this removes one of Cowork's biggest limitations.

Opinions
09 Mar 2026

Claude’s Marketplace Move Shows Where AI Work Is Heading

Anthropic has launched the Claude Marketplace in limited preview. Enterprise customers can now use part of their existing Anthropic commitment to buy selected Claude-powered partner tools. Launch partners include GitLab, Harvey, Lovable, Replit, Rogo and Snowflake. Anthropic handles invoicing for partner spend.