AI Is My Superpower. It's Not My Ghostwriter.
How I Use AI to Write This Newsletter
“Platforms that reward fakeness will create a race to the bottom” —Chris Best, Founder and CEO, Substack
I’m pretty active on LinkedIn. Maybe you’ve noticed.
And, like me, you may have noticed a disturbing trend over the past year or so: a shitstorm of AI-generated dreck produced by “thought leaders” who are prompt-engineering thinly (and sometimes not so thinly) disguised sales pitches to flog products and, hopefully, generate leads.
What bothers me most is that some of it (more than I’d like) reads like posts I’ve written on Substack, with help from Claude and friends.
That’s right, I feel seen.
It’s been nagging at me for a while, to the point where I’ve questioned my own approach, and even whether it’s worth continuing to write in a medium that’s getting hollowed out in real time.
So I was encouraged, honestly, by Substack’s announcement this week about putting AI detectors — aka BS detectors — into the app.
In their own words:
There is an increasing share of text on social media that is generated by AI—as much as 40% on some platforms, according to Pangram’s estimate. This is a big change. Many people complain their feeds are filled with soulless slop. Others don’t yet realize when they are reading something written by no one.
The core problem is not people using AI, or the quality of its output. Not everything made with AI is slop, and not all slop is made with AI. The problem is when there is a mismatch between a reader’s expectation and reality, especially when they unwittingly invest their attention in something with no human thought on the other end. That’s Claudefishing.
I've been writing this newsletter for about two years now, with AI in the loop most of the time, and I've learned — and am still learning — how to use the technology without handing it the keys to vomit out a newsletter under the G2 Weather Intelligence name.
The Wand Doesn't Make the Magician
An AI detection score, whatever number it lands on, is asking the wrong thing. The real question isn’t whether AI touched the words. It’s where the thesis underneath them actually came from.
AI can produce a fluent, confident, well-organized paragraph about almost any subject on earth. What it doesn’t have is thirty-plus years of sitting across the table from retail, CPG, and healthcare decision-makers, learning which weather signals actually move a P&L and which are just noise, where the real complexity lives in turning a forecast into a staffing call or a markdown decision, and how much upside is sitting there unclaimed for the organizations willing to act on it.
Take that away, and what comes back isn’t insight. It’s algorithmic hand-waving — polished, plausible … and empty.
How I Use AI to Write G2 Weather Intelligence
AI knows a lot, but it doesn’t (yet) know what I know. What’s in my head is what I’m putting into these dispatches — accumulated from working with senior leaders in the retail, CPG, and healthcare industries, and before that, warfighters in the Army and Air Force.
Here’s the process I’ve developed since starting the G2 Weather newsletter —
Start with the thesis. The points I want to make are always my own, written or dictated by me first.
Get an outline, and push back on it hard. An outline that survives untouched usually means I didn’t push hard enough.
Get a first draft.
Edit it by hand, section by section. Rewrite what doesn’t sound like me, cut what does but doesn’t belong. Maybe add a Walsh-ism or two. Maybe even a non-gratuitous four-letter word.
Bring AI back in for line edits and fact-checking. Every claim gets checked against a real source before it goes anywhere near “publish.” That step isn’t decorative — it’s caught a retracted study and a misattributed statistic in recent drafts, the kind of errors that would have quietly cost this newsletter the credibility it runs on.
Final review, start to finish, to make sure the piece holds together as one argument instead of a stitched-together set of good paragraphs.
Add a musical coda (h/t Paul Krugman), if it’s the right kind of post.
Publish.
At no point in that sequence does an AI decide what I believe, what I’m willing to put my name behind, or what gets cut because it’s wrong. That part doesn’t scale, and it isn’t supposed to.
Force Multiplier, Not Author
I’m a raging advocate for AI, not a skeptic of it.
G2 Weather Intelligence exists to make the case that these tools, used systematically and with discipline, are among the most critical instruments we have for building a more resilient society — whether that’s a health plan reaching a member before a heat wave, a retailer adjusting staffing ahead of a snowstorm, or a newsletter that fact-checks itself before it publishes.
None of that argues for less AI. It argues for more of it, applied by people willing to be accountable for what it produces.
That, my friends, is AI’s superpower.
Musical Coda
© 2026 G2 Weather Intelligence. All rights reserved.