I come at this through people, workflows, and friction. That feeling of “why is this more annoying than it should be?” is usually where a build starts.
The tools are AI. The judgment is mine.
What I'm building
Two places. One connected idea.
Sparky is who I talk to. The Lab is where I build. They have different jobs, and a way to pass work between them.
01 / THE PERSONAL AI
Sparky AI.
Someone to think out loud with. A little help with the rest of life.
A personal AI running on hardware in my house. A place for everyday questions, things I want to remember, and ideas that aren't quite ready yet.
Conversation with continuityEveryday chat, with room for an idea to become its own thread.
Memory I can readPlain sentences I can inspect, correct, or forget.
Jobs & remindersAsk for a reminder or recurring brief in the conversation.
A voice & a point of viewTalk naturally, and have someone help shape the thought.
Sparky AI The conversation is home.Sparky Lab The work has a place.
App screenshots use sample conversations to show the experience.
Where they connect
Sometimes the build starts with “hey, I had an idea.”
I don't always sit down in the harness with a tidy brief. Sometimes I just fire a thought at Sparky. She helps me find the shape of it, then writes the brief that gives the work a proper starting point.
An idea might start like this
“I want a little place to keep the things we're planning for the weekend. Something we'll actually use.”
— me, thinking out loud
01
I bring the spark
A half-formed thought. A frustration. A “what if.”
02
Sparky writes the brief
We talk it through. She captures the want, the boundaries, and what a good result looks like.
03
The Lab does the work
The brief goes into the harness. I shape the plan, then we build in manageable slices.
04
The result comes back
The work can return to Sparky as a report. I inspect what we made and decide what's next.
The conversation gives the idea shape. The workshop gives it somewhere to go.
How I build
I bring the want. We work out the how.
I didn't come up through computer science. I came up through workflows, edge cases, and noticing where things fall apart for the person using them.
The harness lets me turn that judgment into working software. The agents type faster than I ever will. I still have to know what good looks like.
A running app is the start. How it feels is the next question.
01
Talk until the idea has shape
Sometimes with Sparky, sometimes right in the Lab. Ask questions, challenge the premise, and figure out what would actually be useful.
02
Make the intent clear
A good brief says what I want, what matters, and what we shouldn't break. The implementation can change.
03
Build a slice. Try it.
Get something running in the harness, use it, and find the friction. Small steps make it easier to change direction.
04
Look at the actual work
Review the changes, run the checks, and inspect the result. A confident answer isn't evidence that the thing works.
Inside the workshop
Less guessing. More “show me.”
The plan before. The changes after. These are the parts of the harness that help me keep my hands on the work.
Before we buildA plan I can read and change.After we buildThe actual changes, right there.
Say hello
Building something a little weird, too?
I'd like to hear about it. Personal AI, better tools, or an idea you can't quite leave alone — let's compare notes.
