Operating notes
Practical articles for service SMBs: where to start, what to avoid, and how to get your team actually using it.
Featured read
Mandating use gets compliance, not change. Seven practices for building real capability with a team that pushes back, plus four questions to check your progress at quarter end.
The full archive
Four conditions decide whether AI works in a service SMB: a problem that hurts, someone accountable, information to work with, and team time to adjust.
Starting with the tool is the most common mistake. This is the order that works, and why the step almost nobody plans is the one that decides whether your first AI project gets used.
It is almost never technical. The eight fears that hold back an AI purchase at a service SMB, the control that answers each one, and the signals of a vendor without judgment.
Your team uses AI every day and the business looks the same. Where that gap between adoption and value comes from, and what to check first in a service SMB.
The tools cost the same for a company of fifteen as for one of a thousand. What separates them is who filters the use cases, what information is organized, and who supports the team.
An agent is a language model with connected tools, permissions, and a loop that retries until it calls a task done. That mechanism sets both its reach and its limits.
