Not terrible is a good place to start
When someone says the thing is not terrible, they are usually ready to tell you how to make it better.
From the workbench
A few things I'm thinking about while building proofs of concept (POCs) and talking with people about their work. Questions, mostly, with room to change my mind.
When someone says the thing is not terrible, they are usually ready to tell you how to make it better.
An open chat window feels flexible, but a sorted list and a clear button usually get the job done faster.
Walking through it the day before almost always changes what I plan to show.
Changing a prototype during a presentation feels responsive, but it usually shifts the focus to troubleshooting.
Choose a next step that answers a question, has an owner, and is small enough to complete.
Building something with an AI agent is quick. Finding out if anyone else can use it is the slower, more honest test.
Data quality depends on shared definitions, clear ownership, and a way to correct mistakes.
Many of the people I meet are on their second attempt. What happened the first time is the best place to start.
Telling someone the thing I came to talk about will not fix their problem is uncomfortable, and usually worth it.
A proof of concept (POC) gives people something to question. It does not settle every delivery decision.
Unpaid work is where the limits blur fastest, so it needs them written down more than paid work does.
When a number feels too high, the useful argument is about what is unknown, not about the number.
A description of a system is not the system. Before I size the work, I want to see one real piece of it.