The pressure to "do something with AI" often leads to a flashy demo that never touches real work. Start dull instead: pick one hated, frequent, repeatable task and test whether improving it is worth a week.
Notes from the field,
not the brochure.
Practical notes on using AI in a real small business. No hype, no jargon. Just what works, what doesn't, and what we'd tell a friend who runs the place.
The working archive.
Short, specific reads for the person actually deciding what to do next. Written from real engagements, not a content calendar.
What a useful AI pilot looks like
A good pilot is small, time-boxed, and tied to a number you already track. We share the one-page brief we use to keep pilots honest — and the exit criteria that let you stop one without it feeling like failure.
Measuring time saved without fooling yourself
Hours saved is the easiest metric to inflate and the hardest to defend. How to baseline the before, count the after, and report a number you would stand behind in front of your CFO.
When not to use AI
Every vendor will tell you where AI fits. Almost no one will tell you where it doesn't — which is strange, because knowing when to walk away is most of what makes the technology pay off. A short, honest list of the places we tell clients to leave it alone.
Build vs. buy for a 20-person team
You do not have a platform team, and you should not pretend you do. A plain framework for deciding when an off-the-shelf tool wins and when a small custom build actually pays for itself.
Keeping your data out of the model
You can get real value from AI without handing your customer records to a vendor for training. A practical look at retention settings, data boundaries, and the questions to ask before anything sensitive goes near a prompt.
What an AI agent actually is — and the three safe places to start
"Agent" is the word of the year, and it's doing a lot of work. Strip away the mystique and an agent is just software that can take a few steps on its own toward a goal you set. The question that matters isn't how clever it is — it's where you'd let one off the leash, and where you wouldn't.
The boring work that makes AI worth it
Clean inputs, clear ownership, a place for the output to land. The unglamorous groundwork is what separates a tool people use from a tool people quietly abandon.
Why most automations fail in month two
The build works. The launch goes fine. Then reality drifts — a form changes, an edge case shows up, nobody owns it. Here is what keeps an automation alive past the honeymoon.
The handoffs where work goes to die
When a job stalls, everyone blames a person. Look closer and the delay almost never lives inside the work — it lives in the gaps between steps, where something sits in an inbox waiting to be noticed. The biggest wins from automation aren't faster tasks. They're handoffs that stop dropping things.
The follow-ups falling through the cracks
Most small businesses don't lose customers to a competitor. They lose them to silence — a quote nobody chased, a "let me think about it" nobody circled back on. The fix isn't a bigger CRM. It's making the follow-up happen without depending on someone remembering.
Using AI for content without sounding like everyone else
AI can fill a content calendar in an afternoon. The problem is it fills everyone's the same way — the same tidy, weightless, faintly robotic paragraphs. The win isn't generating more content. It's getting the speed of AI drafting while keeping the one thing that makes content work: sounding like you.
Show us where the work gets stuck.
Bring the repetitive handoff, follow-up, or admin task nobody wants to own. We'll help find the first useful build.
Book a free call
Get Field Notes
in your inbox.
One practical note when we have something worth sending — never on a schedule for its own sake. No pitches, no funnel. Unsubscribe in one click.