Substack Notes — queued for publishing
Notes ready to post. Target cadence: 2–3 per week. Copy-paste one at a time into Substack Notes. Each is standalone. Several are linkable to a longer piece — include the link where noted.
Voice: ad register, punchy. Per /humanize, no “however/moreover/furthermore” at paragraph starts, max 1–2 em dashes, vary sentence length. These are already tuned; don’t let a final pass flatten them.
Week 1 — seed the design-system thread
Note 1 — the agency billing inversion
Your design agency should sell you the system — typography, color, rendering rules — and let you produce your own assets against it.
They don’t, because hourly billing pays better when they own the production line.
You don’t demand it, because “design system” sounds soft in a budget meeting. Much easier to expense “a one-pager.”
Both sides are losing to this. Canva fills the gap, badly.
Link to: Stop paying your designer for one-pagers (Substack article)
Note 2 — the 1841 analogy
Whitworth figured this out in 1841. Every factory cutting its own bolts meant nothing fit together and every repair required the original manufacturer.
Your firm’s design function, most days, is still running on the pre-Whitworth model. Paying the designer per asset instead of per system.
Thread standards aren’t glamorous. They’re what made industrial civilization possible.
Note 3 — Canva as fast food
Canva is to typography what fast food is to cuisine.
Technically the category. Practically a different species.
You didn’t save money by switching to Canva. You paid for your brand to look like everyone else’s brand. The cost just shifted to the thing you can’t invoice: the firm’s visual identity.
Week 2 — seed the judgment-method thread
Note 4 — the Hippocratic checklist
The surgical checklist isn’t new. It’s the Hippocratic Oath in operational form.
Every time you think you’ve invented a management method, check if it’s actually 2,500 years old. Most of the good ones are.
The novelty is almost always in the problem. The method usually turns out to be ancient.
Note 5 — the consulting reframe
Most management consulting is charging you by the hour for things that should be encoded in a document you already own.
The consultant’s judgment is valuable. What you’re paying for badly is the propagation of that judgment — the method that lets your organization hold it after the consultant leaves.
You could buy the method. Most consultants aren’t structured to sell it.
Note 6 — post-AI contribution
AI is coming for your rote work. Fine.
Your durable contribution was never the execution. It was the method you invented that lets other people — or other machines — execute correctly when you’re not in the room.
That was true before 2023. The AI moment just made it harder to ignore.
Link to: Judgment invents method. Method propagates judgment. (blog essay)
Week 3 — provocations and restacks
Note 7 — the fCMO category error
The fractional CMO market solves the wrong problem.
You don’t need someone part-time running your marketing. You need the method that runs your marketing — positioning, content ops, decision rules — encoded as an operating system your firm owns.
A fCMO who leaves after eighteen months takes the method with them. That’s not a sustainable purchase.
Note 8 — “AI readiness” skepticism
“AI readiness assessment” is a racket.
And yet: the alternative — leaping into adoption without a senior outside read on your operation — is worse.
The problem is not that readiness assessments exist. The problem is that most are priced as the start of a six-figure engagement, so the assessor’s incentives are wrong from minute one.
A paid, scoped, bounded readiness read, with no implementation attached, would be honest. Most aren’t.
Note 9 — restack of Michael (SHP)
[Restack Michael’s “Is Technique a Technology?” post with this comment:]
This is the frame. Read it.
Extending the argument this week on the blog: the human self’s durable contribution is not execution — it’s inventing the method that carries the judgment forward. Hippocrates knew this. Benedict knew this. Pacioli knew this. We keep forgetting.
Publishing notes
- Paste each Note into Substack’s Note composer. Attach links where indicated.
- Space them 2–3 days apart. Don’t dump all three in one day; the algorithm punishes bursts.
- Watch the first two Notes for signal — restacks, comments, new subs. Adjust tone for Week 2+ based on what worked.
- If a Note gets real traction, consider expanding it into a Substack article.
When a debate starts
- Reply within 24 hours. Substack conversations die fast.
- Don’t dunk. Restate the claim, add a sharper example, credit the counter-point if it’s real.
- Restack thoughtful critics — Substack rewards civility, not combativeness.
- If someone prominent restacks you, acknowledge and link back to them in your next piece.
When nothing happens
Expected. 2–3 weeks of consistent posting is the minimum before the algorithm starts showing you. Most of the first-month Notes will get modest reach. That’s fine. The compounding starts in month two.
Don’t adjust the plan in the first two weeks. Adjust in week three or four, based on what actually provoked response.
Nobody takes you aside anymore
Print taught a generation when to stop. What we lose when the machines absorb the constraints that used to form us.
Your AI agents need a water cooler
Coordination is a property of the room, not the org chart. What that means when your coworkers are agents.
On the death of the author and the birth of the detector
Why worrying about AI authorship is lazier, and more prejudiced, than it looks.
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.
Memory is (almost) solved. time is next.
AI can't tell if a memory is two minutes or two weeks old. The fix isn't making models feel time — it's cache invalidation: an as-of stamp on every fact, a clock in the context, and a freshness window for anything volatile.
Did the state change? A simple test for whether work actually happened
Either something exists now that did not exist before, or it does not. A simple test for whether work actually happened, and what changes when you build your systems so they can't record anything else.