PowerUp Coaching — Living Knowledge Base

40% of Your Week Is Going to the Tools. You Probably Haven't Noticed Because You Enjoy It.

5 min read

Topic

The unmeasured cost of maintaining an AI stack, why it behaves worse than ordinary tooling overhead, and the single question that triages it.

Target Reader

A knowledge entrepreneur, consultant, or expert practitioner who has built a real AI setup and enjoys building it. They came to AI to amplify their actual work — clinical, creative, advisory — and are now spending a meaningful fraction of their week on the machinery. They have not measured the fraction, because measuring it would spoil the fun.

The Fear / Frustration / Want / Aspiration

The want is clean: amplify my work, don’t generate more of it. The frustration is discovering you’ve inverted it — that you are now working on the thing that lets you work. Under that sits a genuinely uncomfortable question the reader has been avoiding: how much of what I’ve built was necessary, and how much was just enjoyable?

Before State

Experiences every hour of tinkering as investment. Has no ratio, because the tinkering and the work blur together and both happen in the same window on the same screen. Adds tools, skills, and agents at the margin whenever friction appears, on the reasonable theory that reducing friction is good.

After State

Has a number. Understands the two reasons this tax behaves worse than ordinary tooling overhead — the substrate moves, so maintenance recurs rather than amortizes; and some of what you build is pre-obsolete, absorbed by the next model release. Runs a triage question against every workflow instead of adding at the edges. Distinguishes investment from tax by evidence rather than by feel.

Narrative Arc

Jay Drobez — a neuroscientist and clinician — spends most of a mastermind session describing his AI writing stack, then searches out loud for the right word for how he feels about it. He lands on frustrated, and produces a number: 40% of his time on the tools, 60% on the work. He started with a model and a few markdown files; it grew into an MCP server, an annotation loop, integrations, and a topology diagram he had to draw just to keep track of it. The turn is that none of it was a mistake — every addition was a reasonable response to real friction. The resolution is not “build less” but a triage question: what is the one thing that moves the needle for this process? Everything else is candidate tax.

Core Argument

The maintenance cost of a personal AI stack is invisible precisely to the people paying the most of it, because enthusiasts experience maintenance as investment — and unlike ordinary tooling, this tax recurs rather than amortizes and partly funds work the next model release will make redundant.

Key Evidence / Examples

  • Jay’s 40/60 split, self-reported by a working scientist with competing professional demands, and his own framing: “my aim when I use AI is to amplify what I’m doing and to offload the work, rather than to generate more work.”
  • The growth path — model plus markdown files to MCP server, annotation loop, make.com integrations, and a hand-drawn topology diagram. Every step defensible; the endpoint requires a map.
  • The pre-obsolescence argument in Jay’s words: much of what practitioners are building now “will be integrated in the next set of models, which will make this whole ordeal redundant.”
  • The image-compression episode from the same session as a miniature of the whole problem: Jay spent significant time and tokens generating an image in order to save tokens.
  • Lou’s era framing: “we’re in this pre- era, where you had to sharpen your axes” — an honest acknowledgment that this is where we are, not a solvable defect.

Proposed Structure (5–7 beats)

  1. Open on Jay hunting for the word, finding frustrated, and producing the number. Let the number sit.
  2. Establish the sympathy: this is not someone doing it wrong. It’s someone doing it well enough to notice the cost.
  3. The growth path — trace how a few markdown files become a system that needs a diagram, one defensible decision at a time.
  4. Why this tax is worse than normal tooling overhead: moving substrate (recurring, not amortized) and pre-obsolete work (funding a gap that closes on its own).
  5. Why enthusiasts can’t see it: the tax is invisible when you enjoy paying it, and nothing in the experience distinguishes investment from indulgence.
  6. The measurement — one week, two columns, no optimizing while measuring. Then the triage question.
  7. Close on the stake that makes this more than self-care: clients paying for outcomes will not absorb this tax, and if you’ve priced your promise off your own tolerance for a bad week, you’ve quoted a number they can’t reach.

Editorial Notes

The tone problem is the whole problem here. This audience likes building, and a piece that scolds them will bounce. Write it as an accounting exercise, not a lecture — the reader should finish feeling informed about their own ratio, not judged for it. Jay’s credibility is load-bearing: a neuroscientist with real clinical work competing for the same hours is far more persuasive than a productivity commentator, so keep his profession visible. Do not promise the tax goes away; Lou’s honest read is that it won’t for a while, and pretending otherwise costs the piece its credibility. The freeze-don’t-delete instruction in the exercise matters — it lowers the cost of trying, which is why people will actually do it.

Next Step

  • Approved for drafting
  • Needs revision
  • Deprioritised

Graph View