L'écrivain à 30 millions : bâtir un cerveau numérique contre le contenu générique

Why SaaS Lost Its Moat and Why Point of View Is the New One

The last two years pushed a lazy take: AI copies everything, so SaaS has no moat. Nicolas Cole answers with a sharper frame. Any product without a point of view is a commodity — a bundle of features that does a job. A product with a point of view teaches a method. Think of a cooking studio. Giving everyone the same knife and pan creates no edge; teaching a specific way to cook does. People stay for the method, not the metal.

TypeShare makes that distinction concrete. Planning and writing tools are plentiful, yet Cole says the aim was never to sell another scheduler. The promise is to help people execute a particular way of writing online. The split between commodity, personality, and original content is not a taxonomy exercise; it is leverage design for a personal language model . Each kind tunes a different muscle. A copier must keep glancing back at the method owner to decide what a feature even means, so the copier stays steps behind.

Cole applies the same lens to the vibe-coding wave. Shipping an app got cheaper, but getting seen got costlier. Whether you build solo or with a team, a product without a story drowns in discovery. Point of view is a moat in the product and in the writing that carries it. If a launch cannot name the problem, the order of moves, and the language it will own, it remains a feature list. Writing should shape product choices before marketing polishes them.

Language as Open Source and Why Commodity Writing Vanishes

Cole then treats language as open source. The alphabet, common words, and familiar sentence shapes belong to no one. Saying eat healthy to lose weight is a commodity sentence because millions have shaped the same string. When models can spin infinite text at will, the first question is not how to write more, but what is worth writing at all. Anything not attributable to a person drifts toward zero value.

Commodity content is invisible for that reason. A reader who meets a generic wellness tip files it under already known. A model can reproduce the average sentence flawlessly, yet the average is forgettable. Cole speaks from a base of more than 50 million words over fifteen years and admits a tension: repeating the basics for beginners feels dull to him now, while for a beginner that same basic is the most useful first step. Value is always relative to where the audience stands.

An analogy helps. In open-source software, anyone can compile raw code, but the win goes to whoever packages it for a specific job and maintains it. Raw sentences are shared; value appears when you wrap a sentence in a specific lived story, a clear claim, and a next step that fits the reader. Without that wrapper, text reads like a glossary and slips out of memory.

I-Language Versus You-Should Language and Why Personality Wins

The most actionable split in the talk is I versus you should . You should tends to slide into commodity because it repeats what everyone says everyone ought to do. I language stays personal: I tried this, I think this, this worked for me. The first is general advice, the second is personality content, and personality is retrievable. A reader will remember Greg saying I would try this startup angle more readily than a textbook definition.

Greg Isenberg pairs that with two filters for his own work: value and affinity . He first imagines the person in the other seat and looks for a move that helps at that person's current level. Then he trusts that a useful move leaves a trace of goodwill. Hearing that language is open source for the first time links the idea to Cole for a long while, because a new mental handle that helps you act stays attached to its source.

The mechanism runs in four short steps: 1) pick a small lived story from your own recent experience, 2) extract one sharp takeaway, 3) translate it into one doable next step tuned to a beginner or intermediate reader, 4) tell it in I language and dodge preachiness. For example, instead of summarizing a ten-page memo in abstract terms, narrate the single paragraph that blocked you and how you unblocked it. Personality does not scale by copying.

Digital Brain: A Library That Brings Customers Around the Clock

At the center sits the digital brain — a library of your stories, opinions, lessons, and ideas that brings customers around the clock, builds trust, and gives models something real to chew on. Picture a well-labeled workshop. Tools hang in sight and the right one is at hand when the job appears. The digital brain works the same way. Ideas wait on the shelf and jump into posts when needed.

Cole is blunt about timing. Models can now flood the feed, but without raw material they converge on the mean, which the market now labels AI slop. A digital brain supplies raw material that is not mean. Your voice, your examples, and your frameworks keep model outputs distinctive. An empty library plus a strong model still yields a generic summary. A full library plus the same model yields something people can attribute to you.

In practice the brain feeds on three content kinds, each with its own leverage. Commodity builds baseline credibility but does not differentiate, personality builds recall and closeness, original reframes the field by offering a new lens. Cole, who has taught writing online for more than a decade, notes the discipline of honoring levels. The basic step bores the expert yet unlocks the beginner. Respecting that gap is the first planning habit, not an afterthought.

Building a Personal Language Model and Its Leverage Map

A personal language model emerges when you run those kinds with intent. Each new piece trains the system in your phrasing, your examples, and your objections. Over time the assistant stops sounding like a generic writer and starts sounding like you on your best day. Programs such as Ship 30 for 30 , run by Cole and Dickie Bush, exist to fill the library fast. For thirty days you ship one short essay daily, and the habit turns scattered notes into a retrievable asset. Cohorts have run since 2022, with new groups each October.

Cole maps leverage like a small shop. Commodity is the traffic table, personality is the signature product that earns return visits, original is the limited collection that sets pricing power. Competing with models on commodity alone is a losing race because models make commodity faster and cheaper. The path forward is to grow the personality and original layers, where lived experience and a clear method cannot be faked at speed.

A simple three-step setup works: 1) list twenty lived stories from the last ninety days that taught you something, 2) write one sharp opinion and one next step for each story, 3) tag the sixty atoms in a tool like TypeShare and wire them to a weekly ship schedule. TypeShare users then stretch one core atom into tuned variants for different platforms. The model is the accelerator, not the source. Without the sixty atoms, acceleration just makes slop faster.

Language as Data: Million-Dollar Libraries and Brains as a Service

The closing reframe treats a library as data . Corporations buy data estates from distressed firms; buying a writer's full catalog is the same motion. Cole points to deals where teams license estates from carriers such as Spirit for millions. The insight is not the asset alone but the operating muscle around it. The raw grouping of words sits inert until a team can slice, remix, and distribute it at scale across channels.

From that comes a venture idea Cole leaves on the table: brains as a service . One person with a crisp point of view builds the digital brain that the whole marketing engine runs on. Think of a single voice behind Slack or Salesforce or Airbnb that codifies the language and the method, and then a team turns that crystal into hundreds of posts. The cost is not per essay. The cost is building the replicable estate once, then compounding it.

The sponsor note on Brex lands in the same register. Teams at Vercel, OpenAI, and Anthropic are presented as users who moved banking, high-limit cards, and back-office busywork into one stack with AI handling expense reports. Put next to Cole's argument, the picture sharpens. In writing and in finance, the edge goes to the structure that gathers scattered work into one leveraged asset. The digital brain does that for content; Brex is pitched as the counterpart for money movement.