Headless
Headless refers to software that operates without its own visible interface and only provides its content or functions through a technical interface. Other programs, websites, or apps then deliver the presentation that the human ultimately sees.
Most programs consist of two parts. One part is the interface with buttons, menus, and text that you see on the screen. The other part works invisibly in the background: it stores data, calculates, and manages things. Headless describes software in which the visible part is completely missing. What remains is only the machine part, which delivers its data to the outside via a fixed docking point. This docking point is called an interface: a kind of socket that other programs can plug into and retrieve data from.
Why companies build their systems headless
In the past, an online shop was a single program. It managed the products and simultaneously rendered the webpage. That worked fine as long as there was only one webpage. Today, companies sell via the website, via a mobile app, via voice assistants, and via screens in the physical store. A system with a fixed, built-in interface cannot cope with that.
A headless system solves the problem through separation. The product data resides in exactly one place. Every output channel retrieves it from there and displays it however suits that channel. If someone changes a price, they change it once, and all channels immediately show the new value.
There is also a practical advantage for developers. Design and data management can be handled by different teams and with different tools. A new website design no longer means that half the database has to be rebuilt. The downside: you need more building blocks and more specialists, because the interface is no longer supplied along with the system.
The interface as the only exit
A headless system does not talk to humans, but to other programs. The request comes in in a fixed form, for example as a call to an internet address. The response comes back in a format that machines can read easily. Usually this is JSON: a text format in which every value has a name, for example “price: 19.99”.
The program that receives the response is called the frontend. It decides whether the price appears large and red or small and gray. The headless system knows nothing about this. It only delivers the pure information, without any specification for how it should be displayed.
A comparison helps: a restaurant kitchen without a dining room is headless. It keeps cooking, but it does not set any tables. A delivery service, a food truck, and a canteen all pick up the same dishes and each serve them differently. The important distinction: headless does not mean there is no interface at all. It means the interface no longer belongs to the system itself.
Headless software in shops, editorial systems, and AI tools
The term is most commonly encountered in editorial systems. A Headless CMS manages articles, images, and headlines for a news site. The editorial team types their texts into an internal input mask. What the article looks like later is decided by a separate program for the website and another one for the app.
In the retail industry, the same thing is called Headless Commerce. Large retailers use it because they want to redesign their look several times a year without touching the inventory management system. The term therefore regularly appears in stock market and business news whenever providers of such systems announce funding rounds.
The term is also common in the AI world. A Headless Browser is a web browser without a window that loads pages only in memory. Companies use it to automatically collect data from websites or to test software. And many language models are offered headless: the provider supplies only the interface, and the customer builds the chat window themselves.