Skip to main content
Paul Welty, PhD AI, WORK, AND STAYING HUMAN

Work log — 2026-03-12

What shipped today

The knowledge proxy earned its keep today. Three other sessions consulted it via tmux — first the courses session asking about the Maven/Kajabi platform decision, then the supervisor session requesting corpus research for two blog ideas. The proxy answered all three from the corpus, grounding each response in specific documents (DECISIONS.md, platforms.md, the “selective morality” essay, “when the bottleneck shifts to judgment,” “you were trained to suppress yourself,” and more). The tmux-based consultation protocol is now working end-to-end, though we discovered that Enter must be sent as a separate tmux command — chaining it with the text sometimes fails silently.

The bulk of the session was interview preparation. Paul needed corpus-grounded answers to two sets of ten questions each, covering the book’s core thesis as applied to two angles: (1) the machine-self, AI exposure, and what remains human; (2) the coordination tax, permission culture, and designing organizations for judgment. Both sets required deep reads across the book manuscript, course module outlines, blog essays, and voice guidelines. The answers are substantial — each about 300-500 words, grounded in specific book passages and consulting examples, written in Paul’s voice. They live in the new corpus/interviews/ directory.

Along the way, we discovered that all 18 corpus symlinks pointing to the courses project were broken — the project had been renamed from courses to course-work-of-being but the symlinks were never updated. Filed as #397, which was fixed and closed within the session. This restored access to the book manuscript, all six module outlines, audience docs, competitive positioning, LinkedIn strategy, and competitive landscape — the strategic core of the corpus.

Completed

  • #397 — Fix 18 broken corpus symlinks after courses project rename (filed and closed)
  • Created corpus/interviews/ directory with two interview answer files
  • Answered three tmux-routed knowledge proxy queries from other sessions
  • Saved feedback memory: tmux Enter must be a separate command

Carry-over

  • The interview answers are drafts for Paul’s review. He may want to revise voice, add specific client details, or adjust emphasis before using them.
  • The corpus/interviews/ directory isn’t yet indexed in KNOWLEDGE-INDEX.md — should be added if it becomes a permanent corpus location.

Risks

  • The corpus symlinks are absolute paths to ~/Projects/course-work-of-being/. If that project moves again, the same breakage will recur. A more resilient approach (relative paths, or a sync script) would prevent this class of problem.

Flags and watch-outs

  • The book manuscript file has extensive blank lines at the top (lines 1-50 are mostly empty). Not a functional issue but notable.
  • Several corpus subdirectories use symlinks to project files — the courses ones are fixed, but other projects (business, projects) should be audited if those source repos move.

Next session

  • Review interview answer files in BBEdit — Paul may want to add personal anecdotes, adjust consulting examples, or sharpen specific points
  • Add corpus/interviews/ to KNOWLEDGE-INDEX.md if the directory is being kept
  • Audit remaining corpus symlinks for other projects that may have moved
  • Consider whether the blog ideas surfaced today (“coordination as symptom of suppressed judgment” and “AI job apocalypse as mirror”) should become actual blog posts via /blog

Why customer tools are organized wrong

This article reveals a fundamental flaw in how customer support tools are designed—organizing by interaction type instead of by customer—and explains why this fragmentation wastes time and obscures the full picture you need to help users effectively.

Infrastructure shapes thought

The tools you build determine what kinds of thinking become possible. On infrastructure, friction, and building deliberately for thought rather than just throughput.

Server-side dashboard architecture: Why moving data fetching off the browser changes everything

How choosing server-side rendering solved security, CORS, and credential management problems I didn't know I had.

The work of being available now

A book on AI, judgment, and staying human at work.

The practice of work in progress

Practical essays on how work actually gets done.

The machine is eating faster than you can feed it

Sixty-three issues closed across thirteen projects in one day. Four milestones completed. And the hardest problem wasn't building — it was keeping up with what you've already built.

The proxy problem

Every organization has this problem: knowledge locked inside one person's head. Today I accidentally designed a solution — and it has nothing to do with documentation.

True 1-to-1 outreach is finally possible with AI

The 1-to-1 personalization promise is thirty years old. It never worked because understanding each person was too expensive. AI changed the economics.