Your ethics, your vision, your expertise — all valuable.
But only if you express them.
Start. Speak. Write. Articulate.
AI is already listening.
What values are you encoding in your AI work?
The difference?
Who took the time to express themselves first.
I work with AI every day. What fascinates me most isn't what the tech can do.
It's how many good people are still saying nothing.
If you don't talk to AI, someone else will.
And they won't have your values.
We think the AI race is about technology.
More power. More parameters. More benchmarks.
It's actually a race of expression. 🧵
I believe in humanity augmented by tech.
Not humanity absorbed by tech.
The AI tools I build should make people more human, not less.
Closer to nature, not further from it.
No AI could have generated that.
Because no AI has my life, my memories, my weird connections between things.
The most valuable skill in the AI age isn't prompting.
It's thinking.
And thinking requires silence.
I work with AI daily. I build systems that think, plan, and execute.
But I can't automate this: the moment of clarity that comes from doing nothing productive.
The dishes gave me 5 post ideas, a business angle, and a conviction about francophone Africa.
My son was falling asleep. I was doing the dishes.
And I had the most important ideas of my week.
I couldn't type. Couldn't talk. Just stood at the sink, brain on fire.
🧵
The new division of labor:
Humans: create playbooks, inject taste, curate quality
AI: execute at scale, compound learnings
The bottleneck was never AI capability. It was human curation.
Capture judgment. Feed it back. That's how systems compound.
2 years ago this would have cost thousands in servers.
Today a $2000 Mac Mini does the job.
Local tech is becoming accessible.
The real question is: what are you doing with it?
In 2026, anyone can run a personal AI at home.
On your network. With your data. No cloud. No subscription.
My Discord bot responds in 2 seconds, knows my business context, accesses my files, and works while I sleep.
Migrated my personal AI from a Raspberry Pi to a Mac Mini M4.
Before: 1.7GB RAM, 3-10s response time
After: 64GB RAM, 2-4s response time
Haven't even started running local models yet.
🧵
I've been testing frontier models for 3 years. Seen crazy things.
But this was different.
This isn't "AI helps me" anymore.
This is "AI executes while I think."
30 min of my thinking → 10 min of senior engineer work.
Last night I watched an AI plan, delegate, and deliver a feature in 10 minutes.
I touched nothing.
Claude Opus 4.6 + agent teams. Here's what happened 🧵
