Blog category
AI field notes.
Notes from the people doing the work, on what stalls and what to build next.
What these notes cover
Arguments, not announcements.
Field notes take a position and explain it. Read them to test your own thinking before you commit a team or a quarter to an AI project.
-
Why pilots stall.
Most pilots prove a demo and never touch revenue. The pilots note names the missing pieces, usually an owner and a number to move.
-
What delay costs.
The cost-of-waiting note looks at AI decisions the way a finance lead would, comparing the expense of acting with the expense of doing nothing.
-
Build, buy or wait.
The build-versus-buy note decides from the workflow rather than the vendor list, including when waiting is the right call.
Articles
Field Notes: 6 notes, newest first.
-
Oct 4, 2026 · Updated Oct 5, 2026
How to Budget for Your First AI Agent
If the quotes for a first agent do not describe the same job, this note is how to size the work before you agree to a build.
- 10 min read
- Field Notes
-
Sep 26, 2026
AI Agency Pricing: What You'll Actually Pay to Build an Agent
The question is what an agency quote for an agent actually includes, so you can compare the job and not just the label.
- 9 min read
- Field Notes
-
Sep 19, 2026
The Hidden Costs of AI Agents Nobody Quotes You
When the quote covers the build and not the week after, the question is which costs show up once the agent is actually running.
- 11 min read
- Field Notes
-
Jan 31, 2026 · Updated Oct 5, 2026
Build vs Buy AI: Decide From The Workflow
The question is whether this workflow should be built, bought, or left as it is, and the answer starts from the work.
- 4 min read
- Field Notes
-
Jan 17, 2026
The Real Cost of Waiting on AI: A CFO's Perspective
If waiting feels free, the question is what that same delay already costs in the work you do every week.
- 3 min read
- Field Notes
-
Jan 3, 2026
Why AI Pilots Fail Before They Reach Revenue
If a pilot looked fine in a demo and never reached revenue, this note is the break that usually comes first.
- 4 min read
- Field Notes
Questions answered
Field Notes questions.
-
Why do so many AI pilots fail?
They are set up to impress rather than to change a number. Without an owner, a baseline and one workflow to improve, a pilot ends as a slide deck. The pilots note covers how to set one up differently.
-
Is it risky to wait on AI?
Waiting has a cost when competitors answer buyers faster or run leaner, and a benefit when your data and processes are not ready. The cost-of-waiting note helps weigh the two for your situation.
-
Can Claude run parts of a business?
It can run defined routines, such as drafting, summarizing and preparing reports, when the instructions, files and review steps are written down. The operating system note shows how that setup works.
Get AI Systems Notes Delivered Weekly
Get practical notes on AI agents, workflow design, business memory, team routines, and the systems worth building.
No spam. Unsubscribe with one click.
Next step
Test these arguments against your own business.
Our free plan applies the same thinking to your workflows and tells you whether to build now, clean up first or wait.