An LLM With a Logo
Pull ten artifacts from one company in 2026: the cold email, the proposal, the pitch deck, the homepage, the case study, the LinkedIn post, the press release, the white paper, the renewal letter, the follow-up note.
Read them in order. They sound like ten different companies.
I spent ten minutes on a vendor’s site last week, reading their homepage, their case study, and their LinkedIn page back to back. None of them sounded like the same company. The vendor was good, the artifacts were each fine, and the brand was missing.
The email doesn’t echo the deck
Watch how the pieces connect. The email doesn’t echo the deck, the deck doesn’t echo the website, and the proposal doesn’t continue the discovery conversation. The case study reads like a press release. The LinkedIn post reads like a SaaS blog from 2019.
Each piece is fine on its own. Together they’re a stranger introducing himself to you ten times.
The signals are small and consistent. A homepage that says “we partner with our customers” and a sales email three days later that says “circle back on our last touchpoint.” A case study that uses “transformation” five times next to a one-pager that uses “modernization” five times. A founder LinkedIn post that’s casual and pointed, and the same week’s company post that reads as wallpaper. None of these are wrong in isolation, but together they read as a company that doesn’t quite know who it is, and the buyer feels it three or four touches in.
Voice is about one piece. Coherence is about ten
Essay one’s problem was a single piece sounding like nobody. This problem is bigger, because the connective tissue between pieces is missing.
The customer experiences the company as fragmented even when no single artifact is bad. Each interaction starts from scratch, with no recognition and no familiarity building up. The second touch never feels like it’s talking to the same brain that wrote the first one. By the time a buyer has seen six pieces and still can’t describe the company in a sentence, the company has lost the room.
Voice failures get caught because they read as off. Coherence failures don’t, because each piece reads as polished. The company looks correct everywhere and recognizable nowhere.
AI defaults are per-session. Brand is cross-session
Every prompt starts cold. The model has no memory of last week’s deck, and the marketer never wrote down last quarter’s voice rules, if there were any. The AE drafting the proposal has no idea what marketing shipped on the homepage, or what the founder posted on LinkedIn that morning. Each artifact gets the same polished-but-anonymous treatment from the same model, with no shared context between them.
This is what AI-assisted content production looks like at scale in 2026. Every piece passes its own quality bar, and none of them know about each other.
A brand book is not a voice book
Most companies have a brand book covering colors, fonts, logo placement, and acceptable image styles. The whole visual identity has a guardian, a documented standard, and a review process. Nobody ships a marketing piece with the wrong shade of blue.
Almost none of those companies have a voice book: refused words, refused openers, refused sentence shapes, refused closers. The linguistic identity has no guardian, no documented standard, and no review pass. The visual identity gets policed. The voice identity gets whatever the model produces today.
A logo at the top of every artifact doesn’t make them feel like one company, and brand colors don’t either. Templates give visual continuity. Coherence requires linguistic continuity. The first sentence of the cold email and the first sentence of the case study should feel like they came from the same writer thinking about the same reader. Most of the time, they don’t.
The fix
One named owner of the voice standard. Not a committee, not a vendor.
A short standard, not a long one. The voice book I use for my own writing is one page. It says what I don’t do and stops there. That’s the whole thing, and anyone writing as the brand reads it before they ship.
Apply the standard everywhere the company writes: sales emails, AE follow-ups, recruiting outreach, internal all-hands decks. The brand is whatever the customer sees, not whatever marketing produces. The recruiter speaks for the brand, so does the AE, so does the customer success manager, and none of them currently read from the same voice book because there isn’t one.
Review every artifact for voice the way code gets reviewed for security: a named owner who can stop a shipment, a documented standard, a review pass before publish. Essay one made that case for a single piece. Across ten artifact types the wall matters ten times as much, and it still has no owner.
The room from essay one comes back
This is the second time the CIO and the CMO need to be in the same room. Essay one asked whether AI drafts should ship without a human pass. This one asks whether what it ships hangs together, and that question belongs to both of them, not to marketing alone.
The bet
Companies that solve cross-artifact coherence will read as coherent companies. Customers will recognize them after the third piece, the fifth, the tenth. Familiarity will compound, and the brand will start feeling like a person again.
Companies that don’t will read as an LLM with a logo.
A logo at the top is not a through-line.
Brian Beals leads the AI, Analytics & Automation practice at LRS IT Solutions. Reach him at brian@brianbeals.com.