CX Coworker: What marketing teams need to know

Adobe CX Coworker is coming to Workfront, and I think it could make quite a difference to the way people actually use Workfront day to day.

But what is it, what can it actually do, and what should Workfront teams be thinking about before they start using it?

What is CX Coworker?

The easiest way to think about CX Coworker is as the next step on from Adobe's AI Assistant.

Rather than simply asking AI a question, you'll be able to tell Coworker what you're looking for or what you want to do using normal, everyday language.

For example, you might ask it to find a project, show you what's overdue, summarise what's happening across a piece of work or find something buried away in Workfront that would normally take a bit of digging.

And it doesn't stop at finding information.

Coworker can also potentially create and update things in Workfront and work with information held in Workfront Planning.

It's also not just a Workfront feature. Coworker is part of Adobe's wider ecosystem, so the bigger picture is an AI assistant that can help people work across different Adobe applications rather than having to jump between them.

So can it actually change things in Workfront?

Yes, potentially.

And this is the bit I think Workfront admins need to pay particular attention to.

Coworker still follows your existing Workfront permissions. So if somebody doesn't have access to a particular project or piece of information today, using Coworker doesn't suddenly give them access to it.

But there's another important decision for admins.

By default, Coworker can read information in Workfront, but it can't start changing things.

Admins can choose to enable the tools that allow Coworker to write to Workfront too. In other words, creating, updating or deleting information on a user's behalf.

You'll see Adobe refer to these as MCP (Model Context Protocol) tools, but the important bit for most Workfront admins is much simpler:

Do we want Coworker just to find and understand information, or do we also want it to be able to change things?

I wouldn't automatically switch everything on from day one.

What should Workfront admins be thinking about?

I'd start with three things.

1. Are our permissions actually right?

This is probably a good opportunity to check.

Coworker works with the access people already have, so if users currently have access to things they don't really need, it's worth sorting that out before giving AI the ability to act on their behalf.

Good Workfront governance becomes even more important, not less.

2. Where would Coworker actually help our users?

Rather than starting with "What can Coworker do?", I'd turn that around and ask:

"What do our users currently find difficult or time-consuming?"

Maybe it's finding information. Maybe it's understanding where a project is up to. Maybe it's creating work or keeping information updated.

Those are the places I'd start looking for useful Coworker use cases.

There's no point switching on AI just because it's there. It needs to make something genuinely easier.

3. Are people ready for Coworker to take action?

There's quite a difference between asking:

"What's happening with this project?"

and:

"Update this project for me."

Users need to understand when Coworker is simply finding or summarising information and when it's actually making a change in Workfront.

That's as much about helping people get comfortable with a new way of working as it is about the technology itself.

So how would I approach it?

I'd start small.

Choose a group of users who are comfortable with Workfront, find a few genuinely useful things for them to try and see how they get on.

I'd probably start with Coworker's read capabilities first. Let people use it to find information, ask questions and summarise their work.

Then, once you're comfortable with how it's being used, start looking at where allowing Coworker to take action could genuinely save people time.

Most importantly, make sure your permissions, processes and data are in good shape before giving it too much freedom.

I don't think the organisations that get the most out of Coworker will necessarily be the ones that switch everything on first.

I think it'll be the ones that work out where it genuinely helps their users, introduce it sensibly and build from there.