Insights
Insights from production
Writing on what actually breaks, what is worth automating, and how to tell whether any of it worked. Drawn from operating infrastructure where quiet failure is not an acceptable outcome.
Why AI pilots die in month three
The failure is rarely the model. It is that nobody instrumented the thing, so the first upstream change breaks it silently and the outage is measured in weeks.
Read-
What not to automate
Four categories of work where automation reliably costs more than it saves, and the questions that identify them before anyone writes code.
Read -
In most businesses, the integration layer is a person
Every handoff between two systems that cannot talk is staffed by someone retyping. Here is how to find those handoffs and what each one actually costs.
Read -
How to measure whether an automation actually saved anything
If the baseline is not recorded before the build, any number reported afterwards is a guess. A method for measuring that survives scrutiny.
Read -
The pilot worked, and that is not evidence
A pilot runs under conditions production never repeats. Before a good result becomes a decision to scale, four things have to be tested that a pilot is structurally unable to test.
Read -
Where AI earns its place, and where a rule is better
Most workflows described as AI problems are deterministic problems with one ambiguous step inside them. Putting a model where a rule belongs costs money, latency, and the ability to explain what happened.
Read -
The cost of owning an automation after it is built
Build cost gets quoted. Running cost gets discovered. The second one is what determines whether a system is still working in two years, and it can be estimated before anything is built.
Read
Next step
Let's find the workflow worth improving
In a thirty minute conversation we walk through one recurring workflow, the systems involved, and where time, errors, or unnecessary manual work are accumulating. You get a clear assessment of whether integration, automation, AI, or no new technology at all is the right next step.