AI adoption in business: why model power isn’t enough
On 9 July 2026, OpenAI, Anthropic and xAI each shipped a frontier model. On the same day. GPT-5.6 as a lineup (Sol, Terra, Luna), Claude Sonnet 5, Grok 4.5. Three days earlier, Meta dropped Muse Spark 1.1 with a one-million-token context window. We crossed a line without noticing: a frontier model is no longer an event, it’s one more release.
I think we’re looking at the wrong problem. For two years, everyone asked who would have the best model. That question is dying. When three giants release an equivalent-level model on the same day, and a new model lands on average every three days, raw power stops being the point. The real point is elsewhere, and few people talk about it: adoption.
Why model power isn’t enough for adoption
We have a cutting-edge technology, abundant and cheap, and yet the vast majority of freelancers, agencies and SMEs haven’t really adopted it. Not for lack of power, but for lack of a handle. Between an impressive demo and a tool you open every morning there’s a gulf, and that gulf has nothing to do with the size of the model.
Adopting AI requires a clear use case to picture yourself in
This is where the difficulty of the transition lies. For many, AI is still magic, and you don’t picture yourself inside magic. A tool whose concrete gesture and expected result you can’t see, you watch from afar, fascinated or wary, but you don’t fold it into your work. Adoption only happens through a clear use, clear enough that the person sees themselves using it on Monday morning, on their own task.
Three frontier models the same day: the myth of the record
Everyone read that 9 July was “the first day in history” with three simultaneous frontier models. It’s the kind of sentence that makes a good headline and a bad analysis. The truth is simpler: release cycles have tightened so much that a calendar collision is no longer a miracle, it’s become statistics. It’s not a feat, it’s a symptom. And that symptom tells the essential thing: the model is no longer the scarcity. The scarcity is the use that makes you want to get started.
Building with AI: bet on usage and model abstraction
If you build tools, stop chasing the latest model. The one you picked six months ago is already no longer the best value, and the next will arrive before you’ve finished optimising for it. The skill that matters is twofold: abstraction, to switch models without rewriting your stack (this is exactly where a standard like MCP makes sense, it plugs you into the capability), and above all the ability to turn that power into a use someone can picture. The model is fuel. What’s missing are the vehicles people are willing to get into.
Where value shifts when the model becomes a commodity
If the model becomes a commodity, value moves elsewhere: to usage, data and distribution. The moat is no longer in the network weights, it’s in what you do with them and in the data only you own. The labs have understood this, which is why they’re all racing toward agents and “computer use”: the bare model no longer defends itself, usage does. For those who build business tools, that’s good news. The hardest part is no longer technical, it’s human: making the thing concrete enough to picture yourself in it.
From the race for models to the challenge of adoption
We may be living, in real time, the moment when frontier AI stops being a trophy and becomes a raw material. The giants will keep releasing models every three days; it hardly matters anymore. The question is no longer “which model”, nor even “what do you build with it”, but “who, on the other side, can finally picture themselves in it”.
Read next: The UCP protocol, six months on: what the data already says about agentic commerce · Generative Engine Optimization (GEO): how AI chooses the sources it cites · AI agents in business: from the demo to real use
Thomas Nedjar has spent fourteen years in digital acquisition and e-commerce. An entrepreneur, he founded and ran several e-commerce companies between France and Switzerland, with solid experience in international and intra-EU trade. He is now Senior SEO/GEO, AI & Tech Expert at Suisseo, and builds his own marketing applications: ED (automated community management) and SEO Cartograph (local SEO audit crawler). A speaker (CVCI) and trainer, he shares his hands-on insights on SEO, GEO, automation and e-commerce here.
All articles · LinkedIn