What does success look like? Often it might be associated with reaching a certain numeric goal. Now, a problem with this kind of thinking is that we may forget what the actual reason for chasing a specific number was supposed to be. Even worse, the number may not turn out to be a meaningful representation of success in the end.

A new favorite quote of mine that I came across recently is this:

I was reminded about this piece of wisdom when consuming the latest Microsoft marketing content from Build 2025 conference. It included a big number that was repeated in every channel:

Wow, a billion things! Sounds amazing, right? And it’s not just 1 billion, it is 1.3 billion! Which is bigger than the previous 1B figure thrown around:

The source of both these big numbers is IDC. The agents one is from “IDC Info Snapshot, sponsored by Microsoft, 1.3 Billion AI Agents by 2028, #US53361825 and May 2025.” Whereas the earlier IDC Market Note on 1B New Logical Applications didn’t seem to have a visible sponsor, this one was paid for by MS specifically.

Big numbers equal big business - or at least that’s the correlation companies want to make their audience think about. The numbers don’t need to have anything to do about revenue as they seem to work equally well when representing a big chunk of money spent on something. Like the data center investments made by hyperscalers to prepare for AI computing needs of the future.

Creative misinterpretations

Before reaching a billion, you first have to hit a million. Which is a number that Microsoft was able to promote already in their FY25 Q3 earnings call one month ago:

I already pointed out the dirty details of this statement in an earlier newsletter issue. However, I started to see several LinkedIn posts leveraging this specific “million” as a number for Copilot Studio agents. Meaning I now had to point out in detail why this is wildly incorrect.

As a part of this exercise, I tested how easy it is to accidentally create SharePoint agents. Just like I had remembered, the UI of a SharePoint site today is wired to get the user to provision a new agent with as little clicks as possible. This meant that I ended up accidentally creating a duplicate agent already in this testing process. Because I found it hard to believe that the very first click on the “New - Agent” button on the site actions committed the new .agent into the site library. But of course it did.

How about UX for removing an accidentally created agent? That’s a well hidden path. Because it would all go against the primary target that Microsoft has for the feature: the world needs to create as many things called “agent” as possible.

It’s not a novel approach, of course. If we think about the many ways in which the Microsoft cloud suite has encouraged users to create apps and flows before, the idea is exactly the same. When you’re in the early stages of pushing out a new concept into the market, you primarily care about people trying it out. Adoption and active usage comes much later.

Subscribe to Plus to read the rest.