Notes from
inside the build.
What I am learning while shipping. Adaptability, where AI earns its place, and the parts that are harder than they look.
Did Your Last Test Form a Hypothesis, or Validate One?
Most of us would say validate. That's usually the problem. A backward read of six months of data is a hypothesis generator. A pre-designed test is a validator. Most of what we call a "learning" is the first one being treated like the second.
Build Something Nobody Asked For
A colleague built a working agentic platform over a weekend, while playing golf, and took it straight to a live domain. Nothing was riding on it. That is exactly why he tried the things he had never done before.
Free Idea: The Drive That Cleans Itself
Nobody has a clean shared drive. Not the Fortune 100s, not the startups, not me. Here is the tool I wish existed every day, why the two that sort of exist don't get used, and what the real version would have to do.
Not Everything Needs to Be a Deck
Most quick asks don't need slides. Too often it's a 50/50 split or worse between thinking and building. Short docs move faster, invite collaboration, and keep the focus on the work, not the polish.
Your Inbox Is About to Become Someone Else's Problem
We're heading toward a world where we spend more time talking to our assistant about email than actually being in our inbox. And that shift doesn't stop at email.
I Built an AI Wedding Platform Because I Couldn't Picture My Own Wedding
My fiancee could describe every detail. The warm candlelight. The greenery cascading off the tables. I saw an empty room. That gap is where WedVis started.
Speed Is Not the Skill
"Just prompt it." That advice makes it sound way simpler than it is. Speed often gets mistaken for simplicity, and traditionally, "fast" has meant "easy" or "cheap." That mindset doesn't hold up here.
Adaptability Is the Durable Advantage
As nice as it is to stay with one model forever, the real skill that's going to separate people isn't mastery of any single tool. It's adaptability.
The Biggest Threat to Large Organizations
The biggest threat to large organizations isn't other large organizations. It's the person you've never heard of, building a full service offering solo, with AI, over a weekend.
If You Haven't Seen the Code Do the Right Thing, It Doesn't Work
I catch myself trusting the tests my LLM writes as proof everything works… only to get surprised when something breaks in production days later.
Got a reaction to one of these?
I read everything. Disagreement welcome.