The Bezos test: A framework for filtering AI hype in IT

In 2012, at an AWS event, Jeff Bezos was asked what's going to change in the next ten years. He said that was the wrong question and instead should be asked what won't change, because that's what a company can safely build a strategy around.

He stated it simply using Amazon's retail business as an example: It's impossible to imagine a customer saying they wish Amazon offered slower delivery or priced the products a little higher. Both of those things are directly measurable.

There's a lesson to be taken from this story that directly helps IT admins.

Today, the market is filled with a plethora of choices and vendors do not make it easy to choose. Even washing machines come with AI-powered something that claims to improve the quality of your life. Choosing a solution that fits your needs is not straightforward.

This is where you can apply Bezos's theory to filter out real value among buzz words.

What do you, as an IT admin, never say no to? The answer is three measurable outcomes:

• Faster mean time to remediate. The only thing that truly matters is how fast drift, issues, compliance gaps, and vulnerabilities are resolved.
• Less IT spend, whether it is manpower or budget, for more output.
• Better endpoint uptime for business continuity.

This is the filter. Any feature someone is marketing, AI included, only counts if it solves at least one of these three, or it isn't adding any operational value.

Apply any category of endpoint management solution; it has been trying to solve one of the three things.

In the past decade, the market has gone by multiple names. Every few years, the market renames the solution because it either expands to address multiple problems or the underlying technology evolves.

The entire evolution of endpoint management is just successive attempts at the same three outcomes.

• Legacy endpoint management relied on AD and GPOs to deploy configurations and scripts, and troubleshoot endpoint issues at scale. Client Management Tools (CMT) tools lowered the barrier to entry for IT technicians through better UX and workflow-driven management, making endpoint operations faster and reducing IT spend.
• Unified endpoint management (UEM) emerged as mobile computing and BYOD expanded the endpoint landscape, consolidating PCs, smartphones, tablets, and personally owned devices under a single management layer. As security and compliance became integral to endpoint management, UEM evolved to focus not just on configuration, but also on maintaining endpoint health and reducing the time to remediate issues.
• Then came digital employee experience (DEX), shifting the focus from simply managing endpoints to ensuring endpoint uptime.

The names and technologies have changed, but the underlying outcomes haven't.

Now for the challenge: In reality, these three don't move together. Improving one of these outcomes can make the others harder to achieve.

If you push for machine-speed response, your testing windows shrink. Although the patch goes out faster and MTTR reduces, user experience can suffer when insufficiently tested changes disrupt endpoint stability. If you push the other way and take your time to ensure patch reliability, user experience improves and there are fewer bad rollouts. But your MTTR takes a hit. And to achieve faster remediation while keeping better uptime, you need more manpower to verify patch quality, coordinate change windows, and handle edge cases. Spend skyrockets.

That's the endpoint operations trilemma: Speed, continuity, and operational effort are constantly pulling against one another. The goal is to improve all three without impacting one another.

Next time a vendor or your own team pitches you an AI feature, ask them these questions:

1. 01

   Does it reduce mean time to remediation, IT spend, or increase uptime? If the answer is no to all three, you already know what you're looking at.

2. 02

   How does the feature improve one without hurting the other two? If the vendor claims that there is zero trade-off, they're either hiding the operational friction or selling you marketing fluff.