Insights
What we have learned building automation, AI and internal systems for real companies. Written for the people making the decision.
When an AI agent is the wrong answer
Not every task improves by putting a model on it. Here are the three signs that a plain automation will do it better, cheaper and more predictably.
The spreadsheet that became a system
Every company has one spreadsheet nobody dares touch. Here is how to tell when it is time to rebuild it — and when it is actually fine as it is.
How to work out whether an automation pays off
The sum is simpler than most people think. Three numbers, one formula, and an honest answer about whether they are the right hours to save.
Local AI: what you can really run on your own machines
When data cannot leave the building, local models are the answer. But not for everything. Here is what works today, what does not, and what hardware it takes.
Why we build on a fixed price
Hourly billing shifts the risk onto the client. A fixed price forces us to understand the job before we say yes. Here is how it works in practice — and what it costs us.
Five signs your reporting costs more than it gives
Reports that get produced but not read are the most common form of wasted work we see. Here is how to spot it.