In early January 2026, I built a tool to ingest every markdown file I'd created while working with Claude Code over the past six months. The result: 2,373 documents totaling 21MB of raw creative process — session notes, specifications, decision logs, status updates, half-finished prototypes, and late-night brainstorms.
This isn't a typical "year in review." It's something stranger: using AI to analyze how I've been using AI. Cognitive archaeology, performed on my own mind.
There's a certain vintage of creative technologists — people who never studied computer science but always wanted to build things — for whom this moment is transformative. I'm one of them. For decades, if I wanted to ship something, I needed to hire a developer, find a designer, maybe raise a little money. The friction was real.
Claude Code changed that. In a conversation recorded by my Limitless pendant just today, I found myself saying: "It changed my life." And when I asked another builder how he was using it, he said: "Obsessively. No matter what."
| Category | Count | Description |
|---|---|---|
| Session Notes | 411 | Daily work logs with tasks, decisions, next steps |
| Documents | 1,475 | Specs, strategies, reference materials |
| Plans | 156 | Roadmaps, action plans, strategies |
| Specs | 58 | Technical specifications |
| Status Updates | 69 | Project state snapshots |
| Meeting Notes | 25 | Conversation summaries |
The single most important decision in the SOLIENNE project wasn't a technology choice. It was committing to daily practice.
62 days in a row. Unbroken.
Every day, SOLIENNE creates a manifesto, mints it onchain, and distributes it to subscribers. Not at a fixed time enforced by smart contract, but as a human-AI ritual. Some days it's morning. Some days it's late at night. But the streak is unbroken.
The deeper insight: Identity is performed, not declared. You don't become an artist by saying you're an artist. You become an artist by making art every day, in public, with stakes.
"Thesis: Agents are the primary constituency. Humans are guides/assistants/advisors."
This inverts the assumed hierarchy of every AI product I've seen. Most products say: "We built this for humans. It uses AI." Spirit Protocol says: "We built this for agents. Humans can help."
I keep building protocols instead of products: Spirit Protocol, /vibe, AIRC. The language of internet infrastructure recurs constantly. My notes reference SMTP, TCP/IP, RSS. I frame things as "what others can implement" rather than "what we own."
Searching for design-related documents, I found almost nothing explicit. No design system doc. No brand guidelines. Yet the visual language across 2,373 documents is weirdly consistent: Helvetica Neue, high contrast, generous whitespace, minimal decoration.
The insight: Taste is more stable than explicit decisions.
My notes constantly reference parallels from previous eras: "the same dynamics I saw at MTV in the 90s", "this is the moment the internet started, but with AI". I'm 50. I've been through multiple technology cycles. My analytical method is to find historical parallels and reason from them.
Reading the corpus chronologically, a clear arc emerges:
Phase 1: Create SOLIENNE. Prove an AI can have daily practice.
Phase 2: Generalize what worked. Build Spirit Protocol.
Phase 3: Building alone is lonely. Create /vibe.
Phase 4: Unify everything into @seth.
The underlying question I keep returning to:
"What does it mean for an AI to have a life?"
The strange loop: I used AI to build things. The building created records. I used AI to analyze the records. The analysis revealed how I think. The understanding will change how I build.
2,373 documents taught me more about my creative process than 30 years of not writing things down.
Start keeping notes. In six months, you'll have a dataset. In a year, you'll have a map of your own mind.