Headless Gamification: What Is Left of the Game When the Screen Disappears

Anyone requesting leave at a large company today logs into the HR system, clicks through a few menus, picks the days and submits the request. In a headless system it goes something like this: the employee writes in Teams that he needs the week after Easter off. An AI agent checks the remaining leave, creates the request and obtains the approval. The HR system keeps working in the background, only nobody sees it any more.

That is exactly what headless means: a piece of software delivers only its substructure and no interface of its own. The substructure consists of data, rules and actions. In the leave example, the data is the leave balance. The rules determine who may take how many days and when, and who approves. The action is the request itself. The interface, meaning screens, menus and dashboards, is called the head in this language. Hence the word: software without a head. Someone else now supplies the head, for example Teams, a chat or an AI agent.

The topic has now reached HR software. Deloitte described such a system as early as 2024. Workday showed its Headless Access in autumn 2026, and the vendor HONO has been advertising the “world’s first Headless HRMS” since May. The reason is simple: AI agents are supposed to operate this software, and agents have no need of screens.

What Headless Brings and What It Costs

The advantage is obvious. Employees no longer have to click through ten systems, they say in one place what they need. Salesforce data, for example, can already be reached from ChatGPT, Claude, WhatsApp or a terminal without anyone opening the application.

The disadvantages only show at second glance. The vendor loses the place where customers notice it every day, and analysts are already asking whether its product loses value as a result. The employee loses the overview, because anyone who only gets an agent’s answer no longer sees what other options there would have been. And much of it is still announcement: experience from the last Zukunft Personal trade fair in Cologne shows that you hear the word constantly and hardly ever see it in use. Many vendors talk about headless while showing off their own interface at the same time.

The Same Points Under a Different Head

The gamification industry has already discovered the word too. By headless gamification the vendor AWorld means an engine in the background that calculates points and levels and unlocks badges. They are then displayed in the customer’s own interfaces.

That is reasonable if you want no foreign widgets in your product. But they remain the same points under a different head. The question only becomes interesting when there is no head left in which the points could be shown. What happens when the employee speaks only to an agent and nobody opens the portal any more?

The Divorce of Decoration and Game Structure

For gamification, headless works like a divorce, because it separates decoration from game structure. By decoration I mean points, badges and leaderboards. By game structure I mean decisions, consequences and progress towards mastery. Both have sat on the same screen until now, which is why people took them for the same thing.

Decoration needs a surface, because status only arises when others see it. A leaderboard in a chat message is merely a table, and a badge without a profile page is merely a message. Game structure, by contrast, lies in the substructure: that is where it says which decisions exist and what follows from them. None of that needs a screen.

I have argued against this decoration for years and, as you can see in almost any employee app, with modest success. What my argument failed to achieve in 18 years may now be accomplished by an architecture decision like headless.

Let me add one thing: points and badges are of course not completely without effect. The meta-analysis by Sailer and Homner found small but measurable effects of gamification on learning in 2020. For behaviour, however, what worked above all was a game story and the combination of competition with collaboration. Leaderboards also work under certain conditions. But they live on their visual design, and that disappears with the headless approach.

Headless Often Shrinks to a Notification

Headless often shrinks to a notification, and there points survive as a reminder, not as status. Duolingo is the best example. In a research paper in 2020 the company described how it optimises its daily reminders, based on around 200 million messages. One template reads: “You’re on fire. Continue your 109 day Spanish streak on Duolingo.” This sentence needs neither stage nor audience. It merely reminds you that you could lose 109 days. To me that is fear of loss and not pride.

Loyalty programmes are hit even harder. To a shopping agent, a point is merely a discount with an expiry date. It sets it against the price, nothing more. What this means for customer loyalty I have described in my article on agentic commerce.

What Survives Without a Screen

I have always defined gamification differently from the way the industry sold it, namely like this: gamification means designing something so that it is like a journey that unfolds before the user over time on the basis of their own decisions, with a path to mastery, so that ideally you come out cleverer than you went in. Merely triggering people into a behaviour is nudging or a reward system. That can be perfectly useful, but for me it is not gamification.

The best model for this is more than 50 years old. Dungeons and Dragons appeared in 1974 and consisted of rules, dice and a game master who sits at the kitchen table and tells the players what follows from their decisions. There were no screens, and yet people still sit at that table of their own free will today.

An AI agent can take on exactly this role of game master. During onboarding it remembers how the new colleague solved a complaint last week. Then it hands her a harder one and tells her what became of her decision. Headless gamification, then, looks not like Candy Crush but like Dungeons and Dragons.

We have already walked this path, still without an agent. In the wine industry every retailer unfortunately offers the classic points and discounts, so we developed a concept for it that replaces both with a journey. Anyone who has drunk one wine sees which is now the next logical one, and only once someone understands a particular kind of wine does cheese come into it. The point is to make complexity conquerable for the user. Therein lies exactly the satisfaction we know from games, sport and hobbies: doing something through which you experience progress and grow on the friction itself. We used the same concept for software use, where the features developed along a progress tree, and in recruiting, where it was about onboarding in the pharmaceutical sector. This works along my definition, but it then does not look like the classic gamification people usually search for. That is why it is perhaps easily overlooked, and exactly why it works subtly and therefore often more sustainably.

That is also the reason this substructure works in the long run, and headless now forces us to build it. Without a screen you have to go beyond the fireworks of pixels, beyond the ‘chocolate on the broccoli’.

The only users on whom points work reliably are machines, incidentally, and even they cheat. In 2016 OpenAI had a learning agent race boats for points. It went round in circles in a lagoon, caught fire and rammed other boats, yet on average scored 20 per cent more points than humans. The gamification of the 2010s unfortunately treated people in just this way, and I called that Bribery Design.

The Substructure Is the Workplace

If headless catches on, the gamification designer will finally work where game designers have always worked: in the substructure. The three words from the leave request fit surprisingly well. The data holds the state of play, meaning what someone has decided and what they can now do. The rules determine which decisions exist and what follows from them. The actions decide what opens up next and what gets harder.

Whoever is responsible for a programme thereby writes the rules by which any interface can tell the story, whether Teams, agent or notification. In concrete terms: write rules so that an agent can read them, set up consequences so that they really change something in the system, and store progress so that any interface can carry on telling it.

What would be left of your programme if an agent had to tell it tomorrow?