Over the past few years, I’ve been studying the subject of headlessness. This approach allows you to be support agnostic and access everything via API.
This is also the case with this site, which uses WordPress to manage my content and generate a site via NextJS. This allows perfect management of content, images and other resources, all without exposing PHP on the front end.
Media
Media sites represent a fine example of the power of site statics generation, whether total or partial. High-traffic sites have long since switched to a form of headlessness, strictly separating content editing and viewing.
The editing part allows editors and journalists to manage their content efficiently, with flexible tools that last over time, without having to worry about page layout, image formats or even small AI-assisted content creation modules.
On the other hand, on the front end, all media are content agnostic: a journalist who has written an article is likely to be distributed everywhere, on all channels. The website, mobile applications, social networks, whatapps groups, notifications… All published, generated from an article accessible via API.
E-commerce
It took a little longer for the e-commerce part of the business to really immerse itself in headlessness. In my opinion, the biggest obstacle is the e-commerce engine. Often based on monolithic architectures, it’s very difficult to derive a flexible solution and very difficult to migrate once your entire business is based on it.
With a certain lack of knowledge, they can be headless without really knowing it. By doing headless with e-commerce solutions not adapted to this architecture. The result is often shaky, omnichannel without really being so…
Artificial intelligence
We’re all familiar with ChatGPT, Gemini, Claude, Le Chat, Groq… but behind these interfaces lies a world where it’s possible to bypass them and consume only APIs. In fact, it’s very interesting to look at the business economics and realize that there’s still a long way to go to democratize headlessness.
We agree that the interfaces are pretty well done, but they cost a small fortune if you want to use them all. If you use the APIs, you’ll be able to benefit from the same models, on a pay-as-you-go basis. Depending on your usage, you’ll be billed more or less according to your actual consumption. And believe me, if you want to spend €20 a month on a model, you’ve got to go for it if you’re looking for productivity assistance.
The price argument is the first reason, but now let’s look at the power of having at hand, not the best model, or the latest, but all of them at your disposal, according to your demand, take the best to meet your tasks. Today, if you don’t keep up to date and observe everything that’s coming out, it’s impossible to follow and test all the models. We’re likely to see the emergence of tools capable of optimizing models according to demand.
This is already more or less the case with leaders such as OpenAI, which transparently uses different models depending on the instructions you give it.