Field Notes — AI & Automation Mini Cases

Five mini cases from production AI and automation work I have shipped. Each one is laid out the same way:

• The case. What problem, who had it, why the obvious fix did not work.
• The numbers. Real metrics — accuracy, latency, cost, time-saved.
• What worked. The decisions that survived contact with production.
• What failed. The attempts that didn’t — usually the more useful part.
• The architecture. A diagram you could brief an engineer with.
• Next-step checklist. How you’d start the same project on Monday.

If you subscribed via the exit-intent popup you get one of these every two weeks, in your inbox. Or just read them here.