Essay
Follow My Journey
Stop using AI. Start working with it.
Most people treat AI like an app. Open it, ask a question, get a mediocre answer, close the tab, decide it's overhyped. I did the same thing for a while.
Then I started treating it like a colleague I was onboarding. Same tool, completely different relationship. Four moves got me from one to the other.
Move 1: Build intentional context.
This is the one that changes everything. You wouldn't hire a sharp new employee, tell them nothing about your business or how you think, then get frustrated when they guess wrong. But that's how most people use AI: asking it like a search engine instead of briefing it like a colleague.
So I gave the AI a persistent sense of who I am: my work, my projects, the way I think, the people I work with, what I'm trying to build. Every new conversation starts there instead of from cold. That context is doing more work for me than the model upgrades.
If you want this for yourself, here's where I started. Paste this prompt into Claude or ChatGPT and answer the questions. The AI will draft a short “about me” and save it to its memory so it's there for every future conversation.
I want to build a short "about me" you can remember and use to help me better in every future conversation. Ask me 5-7 questions, one at a time, covering: my role and situation, what I'm usually trying to accomplish when I bring you in, how I like to be helped (detail level, tone, when to push back), constraints or context that matter, and what I want you to do differently from your default behavior. When we're done, write a 3-5 sentence paragraph that captures it, save it to your memory, and show me what you saved so I can adjust it. If you can't save to memory, give me the paragraph to keep myself. From then on, use it.
Move 2: Pay for it.
The free tier is the demo. The $20/month tier is the product. Better models, longer context, more tokens, more tools. Once I was paying for it I used it like I owned it. Every day, not just when I remembered. That's when the work started compounding.
Move 3: Apply it to your own problems.
Once the context was loaded, I stopped using AI for whatever happened to come up that day. I pointed it at the actual problems I'd been carrying.
That's when the work stopped being demos and started being progress.
Move 4: Make it part of how you work.
AI isn't a thing you use. It's not a gimmick. It's not an app. It's a colleague you work with, and it gets more useful the longer you do.
Once the first three moves clicked, AI stopped being something I tried and became how I worked. Daily. On real problems. The patterns I kept reaching for, I codified into named skills the AI invokes on demand: a time-log that captures any work hours I mention and rolls them into a weekly tally, a commitments tracker that schedules a nudge before any deadline I commit to, sprint reviews, council-style decision pressure-tests. Each one started as a one-off and graduated to a primitive.
AI doesn't replace expertise. It rewards it. The deeper you go in your own work, the more it gives back.
Planetary Pulse
This one wasn't for a client or a project. I just wanted to take a pile of unrelated real-time data and turn it into something you can read at a glance: partly to see if I could, partly to keep testing the process with AI and stretching what I could do with it. I picked the Earth because what's more complex than the Earth?
A slow-rotating Three.js globe pulling live feeds from:
Built with Claude Code. That's really why I'm showing it: you don't need a client or a deadline to start. Pick something you're curious about and go build it.
Close
The difference between AI that frustrates you and AI that changes how you work isn't the model. It's intention, and intention is teachable.
Go. You've got everything you need.
And if you'd rather build it together, that's the work I do now: one working session where we onboard AI to how you actually think and run your work, and you leave working differently the same day.