You're publishing more than you ever have. The pipeline hasn't noticed.

We’re two weeks into applying Kellogg's I-MOS framework (Prof. Mohanbir Sawhney) to content operations. We've done Memory. This week: Modular services — the layer that decides whether AI output sounds like your company or like no company in particular.

Here's the tension every content team is sitting in.

Bespoke is too slow. Hand-crafting every asset from a blank page doesn't scale, and it's the first thing AI was supposed to fix.

But unconstrained is worse. Point a model at a topic with no reference material and it gives you something polished, plausible, and interchangeable with what your three closest competitors published the same week.

Modular is the way out of that trade-off. And it starts with something I call Brand DNA.

Brand DNA is your voice, POV, proof, and rules of engagement — coded once into reusable blocks the AI reads on every session, instead of re-deriving your brand from scratch on every draft.

The difference between explaining who you are every single time, and handing over a reference the engine already holds.

I run ClaritEdge on Claude Pro, with a project per brand. Each project holds six things. None of them are prompts.

1. Custom instructions. Not "write in a professional tone." The standing operating rules: who the audience is, what we're selling this quarter, what the deliverable format is, where outputs go, and the naming convention. This is the layer that stops every session from starting with orientation.

2. The brand book, including positioning and messaging.

3. The brand voice brief. Sentence rhythm, the words I use, the words I'd never say — written as examples and anti-examples, not adjectives. "Calm, no hype" is an instruction nothing can act on. Two before/after sentences are. (This is the same brief I give clients and I’m offering a fillable brand voice brief here.)

4. The POV file. My defensible positions, written as claims I'd argue in a room. Not themes. Claims. Same spine, different surface, whether it lands in a post, an email, or a byline.

5. Proof points. Anonymized results, the objections that come up on every call, the real before/after numbers. This is what makes a claim land instead of float — and it's the file most people never build.

6. Guardrails. The never-say list. Banned framings, claim discipline, the phrasings I've retired because I've overused them. Guardrails are the file that gets updated most often, because every draft I edit teaches it something.

Because the blocks compose.

A launch email, a LinkedIn post, and a webinar abstract can pull the same POV block and the same proof block, each expressed for its format. Consistent, because it's the same DNA. Fast, because nothing gets re-derived. Personalized, because you swap the blocks that should change and keep the ones that shouldn't.

That's the real mechanism behind "personalized at scale" — not a thousand bespoke pieces, and not one template stretched thin. Reusable DNA, assembled per buyer and per moment.

The one shift that made it stick: the DNA lives in files, not in a chat window. I work in Claude's Cowork mode, pointed at the brand's actual folder, so the voice brief and the proof points are read from disk — not pasted in, not remembered from last Tuesday. When I revise something, I update the file. The next session starts from the corrected version. No re-teaching.

That's the difference between a prompt library and a system. A prompt library is instructions. Brand DNA is your brand, in a form the machine can use.

Don't build all six. Pick your single most-repeated POV — the position you find yourself re-explaining in every piece. Write it as two lines: the claim, and one proof point. Save it as a file, not a note.

Paste it into your next three drafts and watch how much less editing they need. That's your baseline. Everything else is the same move, repeated.

Reply “DNA” and I'll send the Brand DNA block template — the six block types with a filled-in example of each, so you can start coding yours this week.

And tell me: what's the one POV you re-explain in every single piece of content? That's your first block, whether you've named it or not.

— Jill

P.S. Next week: Orchestration — designing content journeys for AI-mediated search, so discovery is something you design, not something that happens to you.

Powered by beehiiv.