What does multiplayer AI actually unlock for my business?

Anya AndrewsAnya Andrews
-November 2, 2026
What does multiplayer AI actually unlock for my business?
Stories from the field
As a Solutions Engineer at Dust, I spend a lot of my time selling Dust to people who already use Claude extensively.
They show up to calls with skills they’ve built, workflows they’ve automated, and very reasonable questions about why they would pay for something else. Sometimes they’re internally building the thing we’re discussing while we’re discussing it.
I like these calls. People have been building great tools for themselves! But the conversation gets interesting when we talk about everyone else at their company.
Who else is using what you built? How did they get it? What happens when you change it? And how much of your week are you willing to spend helping your coworkers become slightly worse versions of you?
Because that last part is a job. Possibly not the job you wanted.

Your sellers do not want a new hobby

One enablement leader told me about a seller who had turned Claude into his personal secretary. He was on top of his accounts, knew how to communicate with each customer, and had a setup that worked really well for him.
The rest of the team was not doing that.
Some people weren’t interested in AI. Others weren’t going to follow all the steps required to reproduce his setup. The enablement leader wanted useful information pushed into Slack, where sellers already worked. Ideally without ten buttons and a small configuration project.
He also asked the obvious question: we can already share Claude skills, so how is sharing in Dust different?
Fair. If my answer is “you can send your coworker a skill,” we probably don’t need the rest of the demo.
But in Dust, sharing the instructions is only one part of multiplayer AI. We ask who manages access, keeps the workflow current, and checks whether the team is actually using it?
What he needed was a way to give sellers the benefit of the workflow without giving them responsibility for setting it up. They should be able to get their account research and go into the meeting. An interest in prompt engineering should not be a prerequisite.
Another conversation brought up the question of how to roll a competitor-intelligence workflow built in Cowork out to the whole team. It would research competitors and produce material for reps. But before rolling it out to all the sellers, one question came up:
“Where is this going to live?”
He wanted a central library, approved skills, and ownership. He also did not want everyone creating their own things until the library became another place nobody could find anything.
I appreciate this level of specificity. We’re past “can AI do competitive research?” We’re discussing whether the reps will find the research before they ask him to do it again.

Please update your plugin (again)

Another customer walked us through their skill distribution process.
They published updates through GitHub. Each user then had to update a plugin to get the latest version. Anyone who didn’t was still using the old skill.
To track usage, they were having the skills write activity into a Google Sheet.
I respect the resourcefulness. I also understand why the people maintaining this setup were interested in alternatives.
They wanted their operations team to manage the workflows and their sellers to use them in existing Slack deal rooms. Their champion was very clear that buying Dust should not mean giving AEs another application to visit.
This is a much better explanation of multiplayer than anything involving a stock image of people collaborating around a laptop.
RevOps changes the follow-up process. Sellers get the updated process. RevOps can see whether it’s being used. Nobody has to chase the person who missed the plugin update because they were, inconveniently, selling.

I would also like to stop being the human search engine

This part is personal!
When I run a pilot, there’s the tidy version of what happened, and then there’s everything the next person will eventually ask me.
Why didn’t we build that use case? Who wanted to join? What did we decide about that integration? Was that a hard requirement or something someone mentioned once?
I can write a handoff document. I cannot predict every question my future teammate will have (and if I put absolutely everything in the document, I have created a different problem).
On one customer call, I showed them the Pod I was using to manage their pilot. The conversations and files were together, so someone taking over could ask about the work we’d deferred, rather than relying entirely on my recap.
The customer said they’d known collaboration was a differentiator, but hadn’t really understood how they would use it until seeing that.
That’s the part I care about. Someone else should be able to pick up the account with access to the context behind my decisions. Ideally while I’m doing something other than answering questions about the account I just handed over.

So what would I put in a pilot of Dust?

Something your team already needs. Preferably something a capable, enthusiastic person has already built and is now mildly regretting agreeing to maintain for everyone.
Give it to the people who are supposed to use it. Give it to the people who are supposed to use it. Can a seller use it without help from the builder? Can someone else maintain it? When the process changes, does everyone get the update? Watch where people get stuck, make changes, and check whether it actually helps them do their jobs.
These customer conversations don’t prove every shared workflow is valuable. They do make a pretty good case for testing what happens after you get started building with AI.
I want the person who loves building to keep building. I want the seller who has six calls today to get the right prep without developing a passion for GitHub.
That seems like a reasonable thing to ask of multiplayer AI.
Want to see Dust in action? Let’s connect →