Fable 5 and the Dynamo

When factories first swapped steam for electricity, towards the end of the nineteenth century, something strange happened: nothing. Productivity barely moved for thirty years. The dynamo was a marvel, but factories were still built around the central drive shaft, machines bolted wherever the belts used to run. The gains arrived only when a later generation rebuilt the floor around the technology, instead of fitting the technology into the floor.

We think about this every time a CMO asks whether Claude Fable 5 is marketing's seismic moment.

The model deserves attention. It is the most capable thing Anthropic has released, gated on arrival while its more permissive sibling, Mythos 5, sits behind a velvet rope for vetted organisations. It is priced accordingly. But Fable 5 is too new for adoption data of its own. What we do have is two years of hard evidence about what happens when marketing meets a powerful model, and it tells one consistent story.

Start with the obvious application, content at scale. Coca-Cola has now remade its Christmas advert with generative tools two years running: the 2024 version was widely jeered as soulless, the 2025 one less so, and both were markedly cheaper and faster to make. 

Research in Science Advances found that writers using AI produced individually better but collectively more similar work. The cost of saying things collapses while the cost of being heard rises, because every competitor owns the same motor. A stronger engine doesn't supply a point of view. It multiplies the one you feed it.

The finance picture is largely the same story, an MIT report last summer found that 95 percent of enterprise AI pilots had produced no measurable P&L impact. Klarna, the poster child, cut roughly $10m a year from its marketing costs with generative tools, then spent months walking back parts of its automation after quality slipped. The lesson isn't that the technology fails. It's that a line item with nothing rebuilt around it is exactly what your CFO suspects: an expensive way to feel modern.

The rewiring, when it happens, is mostly human. A landmark study of AI at work found productivity gains of 34 per cent for novices and almost none for experts, which tells you what these systems substitute for: accumulated experience. So when drafts are nearly free and judgement is the scarce input, an awkward question follows. If the model now does the junior work, where do your next senior people learn the craft? You are not simply buying capacity. You are deciding who gets trained.

So, seismic? 

Seismic moments rarely announce themselves on launch day. They show up two budget cycles later, in the org chart and in the P&L. Our challenge to CMOs stands. Stop asking whether your teams use Fable 5; they do, or shortly will. Ask what you have rebuilt around it, and name one result a customer would notice. If you can, you are ahead of 95 per cent of the field.