Field notes
Short essays on practical AI, customer problems and knowledge that doesn’t get lost: the ways I try to make work less messy.
AI is only useful to me if it makes someone’s judgement sharper, not if it makes the decision for them.
What draws me to AI is pretty simple: it’s good at finding patterns in messy information and cutting out the busywork, fast enough that you can actually experiment with how you work.
My default is human in the loop. AI can dig up and organize the information. But someone still has to bring the context, check what actually matters, and own the call.
AI can hand you a decent first draft. A person still has to own the call.
AI can speed up research and hand you a decent first draft. But when the work touches a real customer, a person still needs to own the context and make the final call.
I don’t really care if the output sounds convincing. I care whether I can still see the sources and assumptions behind it well enough to decide for myself.
Compare notes by email
The real question isn’t how fast you can answer. It’s whether you understood what you were answering.
The first symptom is rarely the full story. Ask a few more patient questions and you usually find the real workflow hiding behind the ticket, and often an answer that helps more than one person.
In customer work, a request often arrives as a technical symptom. You only find the useful next step once you understand what the deadline actually is, and what this problem is stopping someone from doing.
Compare notes by email
A bigger archive doesn’t help anyone. Enough context for the next person does.
A resolved case becomes more valuable when the next customer or colleague can benefit from it too. Clear articles and structured knowledge turn individual problem-solving into a team capability.
What helps is enough context and reasoning left behind that the next person can act on it without having to start from scratch.
Compare notes by emailThe questions I keep returning to.
Disagree with one of these, or stuck on a question of your own about customer work or AI? I’d like to hear it.
Let's talk