Can AI turn one article into a class, a newsletter, and a post?

The question

I do a piece of real thinking maybe twice a month. It needs to become four things: a Lab entry here, a section of a class, a newsletter, and something short on LinkedIn. That's not four ideas, it's one idea wearing four outfits.

So: can the outfit-changing be automated without the thing underneath going dead? I've been running this for five months and I want to be careful about overclaiming, because this is the lab where I most wanted a yes.

What I built

A chain, not a button. The source piece goes in, and each output has its own pass with its own prompt and its own set of examples from things I've written that I actually liked. A newsletter pass, a teaching pass, a short-form pass. Each one gets told what the format is for, not just how long it should be.

The teaching pass is the only one that's genuinely good, and I think I know why: turning an argument into a lesson is a structural transformation, and structure is exactly what these tools are good at. Find the claim, find the evidence, find the thing a beginner would get stuck on, put them in order. It does that better than I do at eleven at night.

Where it goes flat

Short form is the worst of the four, every single time. It produces something that reads like content: correct, competent, weightless. It reliably strips the one specific detail that made the original worth reading — the number that surprised me, the thing that went wrong — and replaces it with a generalization. It's optimizing for something. It isn't optimizing for truth.

The newsletter pass is close but always needs the top and the bottom rewritten. It cannot open. Every draft starts with a windup sentence that says nothing, and every one ends by summarizing what I just said as though I'd forgotten.

And there's a failure I didn't anticipate: on a piece where my own thinking was muddy, the output came back clean and confident. That's dangerous. It laundered a half-formed idea into something that looked finished, and I nearly shipped it. Now the first question I ask is whether the source piece was actually done.

What I changed

I stopped treating the outputs as drafts to edit and started treating them as outlines to write from. That single reframe cut the time I spend fighting the text roughly in half, because editing bad-but-fluent prose is slower than writing from a good skeleton. Fluency is the trap.

I also cut the number of outputs from six to four. Two of them existed only because automation made them cheap, which is a bad reason for anything to exist.

Where it stands

Halfway, honestly. It saves me perhaps three hours per piece and it has definitely made me more consistent — this Labs page exists partly because the friction of writing it up dropped. But every single thing that goes out still gets a full human pass, and on current evidence that isn't a temporary condition I'm going to prompt my way out of.

If you're evaluating one of the all-in-one repurposing tools: they will do what I've described here, competently, and the flatness I'm describing is the thing you'll be deciding whether you can live with. For a business whose product is the thinking, I don't think you can.

→ The workflow is taught in the LifeQuest content track

← Back to all labs