Estimated read time: 6 Minutes
Author: Jason Cross
It was Grady Booch who famously said “A fool with a tool is still a fool.” So, when I stumbled across this quote, it reminded me of the early part of my career.
The organisation I worked for at the time had major performance issues with one of their critical applications. The poor performance of this application was costing the organisation money as it directly affected the amount of revenue the organisation was able to generate.
It so happens that we had some Application Performance Management software licences we inherited as part of a larger software deal, which was just lying around gathering dust. I was, (and still am), a huge fan of APM and had already spoken to a variety of colleagues about the benefits of APM.
With the situation becoming increasingly desperate, we got the go-ahead to install the APM tool, with the expectation of the tool telling us exactly what the issue is and how to resolve it. This seemed so easy at the time, but boy was I wrong!
The Tool
The problem was I inherited a tool I didn’t know how to use. Don’t get me wrong, it was a great tool with fantastic coverage, however, I was drowning in metrics.
When you have thousands and thousands of metrics to look at, where do you start?
This situation is not unusual! For example, one of our customers had over 400,000 real-time charts available to them on the operations bridge. So, the team there ended up using the basic metrics, and repeatedly missed important service issues.
Find out if you're getting the most from your cloud and AI cost tools
Get a cloud cost diagnostic
I was also under pressure from the senior leadership team to provide answers quickly, but I didn’t have any, which left me looking like a fool. This is not a situation any of us want to find ourselves in. Worse still, I would go further than Booch and say, “A fool with a tool is a dangerous fool.”
I have come across a few of those in my time and boy do they wreak havoc. Dangerous fools are those who are armed with tools but don’t have the knowledge or insight and make sweeping statements about scaling, performance, cloud cost and AI spend that can cause organisations to waste money, place them at risk and even worse, permanently lose customers.
Finding the important key metrics is often like finding a needle in a haystack. And all too often, a new tool will only make the haystack bigger!
Same trap, new tools
Fast forward to today and the tools have changed, but the trap hasn’t. Where I had an APM tool and thousands of performance metrics, teams now have cloud cost dashboards, FinOps platforms and AI spend trackers, each one splitting the bill by service, tag, model and token.
It looks like control. But a dashboard can tell you the bill went up. It can’t tell you why, who should act, or whether the fix is in the architecture, the model or the way your teams work. And like my APM tool, a new platform can make the haystack bigger.
The numbers say the same. In the Tokenomics Foundation’s State of Tokenomics 2026 survey of 472 companies, three in four couldn’t confidently show their CFO a business outcome from AI. Proving value was the biggest challenge for 43%. Cost and pricing complexity was the biggest for 7%.
And the dangerous fool is still around: “just move to a cheaper model” or “cut the idle capacity”, said without knowing what that capacity is protecting.
How can you avoid becoming a “Fool with a Tool”?
The most important things you need in these situations are processes and methodologies. Most technological issues are complex involving a variety of interlinked systems. So having a process will help guide you and your teams through to find a resolution and without missing anything.
With cloud and AI, that process starts with ownership: who can approve a new AI use case, what triggers a review when a number moves, and who is accountable for the outcome rather than the bill. That is what a FinOps transformation is really about.
The methodology is equally important, as it tells you how you will address the issues in the most cost-effective manner. Look at cost, demand and performance together. Cut cost without watching the other two and you can starve the workload that pays for everything.
By focusing on behaviours, you will be able to scale effectively to handle peaks, without throwing money at the problems. That is as true for a new AI feature that suddenly needs far more compute as it is for a seasonal peak, and it is the heart of controlling AI spend.
I am still a huge fan of the tools. I just wouldn’t hand one over without a process, a methodology and an owner.
Sources: The State of Tokenomics 2026 (Tokenomics Foundation, September 2026).
