Applied AI

Do I use AI? Absolutely.

I also use electricity, indoor plumbing, and a waffle iron. “Human in the loop” is today’s buzzword. My goal is to get ahead of the loop, use it like a hula hoop, with me firmly in the middle, and circle back to make sure it’s doing what I intended.

I turn human needs into usable workflows, clear instructions, checkpoints, and tests. The computer gets a job. It does not get the final say.

Real need → usable workflow → human approval → evidence and read-back.

A workflow I created

Finding what needs attention

I created an AI-assisted workflow that brings relevant information together, helps assess urgency, and keeps a tracker with daily notes and updates. It helps me decide where to focus when the workload is heavy and the stakes are high. I keep the final say.

Making progress visible

I used AI to help build a dashboard that shows what is finished, what still needs work, and who owns the next step. The goal is less chasing updates and more time doing the work.

Connecting ideas to finished work

I use AI to help move letters from idea to website, build practical guides, and make the collection easier to browse and search. Voice, accuracy, and publishing decisions stay with me.

Shaping the help around the person

I designed and prototyped a builder that helps tailor AI support to a person’s needs and preferences. Clear boundaries, human review, and testing are part of making that help useful.

Hands-on exploration

Learning by building

I’m using small, practical projects to understand how applications work. An interactive weather classroom lets me explore how information moves between systems and whether the result matches what I asked for.

Keeping useful context

I’m developing ways to carry approved decisions and project context into the next session. The aim is less repeated explanation, with checks that the system actually follows what we agreed.

Finding the evidence in my own words

I used Copilot to find recurring strengths in years of performance-review feedback, then made a word map I could return to when confidence needed evidence.

See my actual word map or try the exercise yourself.

Checking that “done” means done

I build read-back and review into the process: check what was saved, test links and search, and look at what a visitor actually sees. Today’s website work is a good example. An “OK” message isn’t the same as a working result.

Keep the conversation going