What Claude Fable can do for an expert firm, shown, not claimed

Claims about AI are cheap right now. Every tool is revolutionary, every demo is stunning, and every vendor promises the future.

So I am not going to claim anything. I am going to show you what one trained system produced, with the numbers attached, and let you judge. Then I will tell you plainly what it cannot do, because that half never makes the sales page.

First, what "trained" means

Claude Fable is Anthropic's most capable model, and out of the box it is brilliant and generic. It knows nearly everything and it knows nothing about you.

Training, in my practice, means closing that second gap. I interview a client for three recorded hours, turn the transcripts into a suite of brand guides, and load the guides into a Claude project that is refined until the output is ideal for that one brand: right facts, right reasoning, right voice. The system's ceiling is the depth of that capture, which is why the interviews come before everything.

That is the machine behind each demonstration below.

Shown: a website in twelve working hours

For one client, the website went from approved brand guides to fully built in twelve working hours.

Not twelve weeks of a team trading briefs, and not a template with the client's logo dropped in. Page copy in the client's voice, structure fit to their buyers, built and refined by one person operating the trained system. The calendar time was longer than the working time, because my clients' schedules set the pace. The working time is the point.

Shown: nine carousels promised, twenty delivered

One engagement scoped nine social carousels. I delivered twenty.

Not because I underestimated the work, but because once the system holds the whole brand, volume stops being the expensive part. The expensive part was always the knowing: what the brand sounds like, what its buyers care about, what claim each asset should carry. Captured once, that knowledge feeds every asset, so generosity inside a scope costs the practice little and the standard does not move.

Shown: a complete brand in weeks

A functional medicine practice in a resort town came to me with a coming-soon page. Weeks later it had a complete brand system: foundation, voice, website, content, the working engine.

The compression is the demonstration. Every one of those assets grew from the same three hours of capture, which is why none of them contradicts the others. Consistency at that speed is something I never saw a team of specialists produce, and I say that having worked alongside good ones.

What the demonstrations share

Look at what all three have in common, because it is the part vendors skip.

None of the speed came from the model alone. It came from the capture that preceded it and the refinement that followed it. Every deliverable passed through my hands, and plenty needed them: the system produces, and judgment decides what ships. The craft is in the refinement.

That is also why the client's hours matter so much. The system multiplies what it is given. (The full architecture, stage by stage, is in how one person and a trained system replace a marketing department.)

Not claimed: what Claude Fable cannot do

Now the half that never makes the sales page, because you deserve the whole picture.

It cannot supply experience nobody captured. Untrained, Claude gives you the same competent, generic layer it gives everyone, and generic is exactly what an expert firm cannot afford to publish.

It cannot judge alone. The system drafts to a standard; deciding whether a piece is true, wise, and worth your name stays a human job, and in my practice it stays mine.

It cannot create demand. A trained system builds the presence, the content, and the consistency. Your reputation and your work still earn the attention all of that converts.

And it cannot fix a commitment problem. An owner who will not sit for the capture gets a thin system, no matter whose hands operate it. You will be the bottleneck, and I tell every client so before we begin.

Judge the receipts

Twelve working hours. Twenty carousels where nine were promised. A complete brand in weeks, from three hours of interviews.

Those are the receipts, offered instead of adjectives. Weigh them however you like.

And the question worth taking with you: trained on your experience, in your voice, what would the system produce? The honest answer is that neither of us knows yet, because your version of it does not exist until your experience is captured.

Brett Wharton, looking ahead