DFBW 2026

From Service to System: Dining’s Fork in the Road

Posted by foodarchitectonics on

I work in hospitality. And lately I’ve been watching the “middle” thin out.

A few years ago, guests who weren’t “the target crowd” still came in—curious, occasionally intimidated, but excited. You could feel the moment when someone discovered that a good bar or restaurant isn’t a private club: it’s a public space where taste, warmth, and a small sense of celebration can be shared. Recently, that moment happens less. Not because people stopped caring about food, but because an ordinary night out increasingly feels like a decision with consequences. When participation becomes expensive, dining stops being a social common and starts becoming a marker of class. You can feel it in the small behaviours: fewer “let’s see where the night goes” orders, more careful mental math, more people treating the menu like a risk assessment instead of an invitation.

The pressure points are not mysterious. Costs rise across the board, while many price structures were built in an era where hospitality survived through thin margins and cheap labour. Add the industry’s habit of romantic chaos—intuition without systems, passion without planning—and the result is a sector that is both culturally important and economically fragile. Swiss industry reporting paints a picture of ongoing strain: staffing challenges, uneven performance, and a fragile baseline where small changes in costs or demand can flip a business from “working” to “not working.”
Even consumer prices for categories including “restaurants and hotels” have shown persistent upward pressure in recent years.

This is the environment into which “digital transformation” arrives. And some of it is genuinely useful: better demand forecasting, tighter inventory control, smarter staffing, less waste, fewer mistakes at the point of sale. Self-order kiosks, for example, don’t just reduce friction—they can meaningfully change purchasing behaviour (University of Hamburg reported higher spending at kiosks in two studies).
In other words: tech doesn’t only optimize operations; it can also optimize consumption.

But technology is not a fairy godmother. Even the most hyped “frictionless” concepts have had to adjust once they hit the messy reality of scale, cost, customer expectations, and trust. Amazon, for example, confirmed it would replace “Just Walk Out” in its Amazon Fresh grocery stores with smart carts—still faster than a classic checkout, but with more transparency about spending.
That’s an important signal: the future isn’t simply “more automation.” It’s “automation that survives contact with real humans.”

Now let’s exaggerate—on purpose.

Worst case: hospitality splits cleanly in two.
The affordable tier becomes relentlessly automated and emotionally thin: screens instead of greetings, standardized menus instead of judgment, dining rooms designed for throughput, not warmth. People don’t choose depersonalization; it becomes the default because it’s all they can access. Human service survives, but behind velvet ropes—craft, conversation, and care reserved for those who can pay. Over time, the middle disappears: fewer “normal” places where a broad mix of people can still afford to be served by other people. Food quality becomes quietly more industrial, more anonymous, more engineered for cost efficiency. And the logic ends in a dark joke: Soylent Green not as a literal prediction, but as a metaphor for optimization without dignity.

If that sounds melodramatic, it’s because I’m deliberately borrowing dystopian lenses to make the trajectory visible. The nightmare isn’t killer robots—it’s the soft, polished version: Brave New World comfort as social anaesthesia; WALL·Ehumans as passengers of systems they no longer steer; Black Mirror futures built not with violence, but with frictionless design. In that world, dining becomes an interface: timed seating, algorithmic upsells, “personalization” that is really just conversion optimization. And the truly unsettling part is that the building blocks already exist in friendly forms—vending-machine hot meals, conveyor systems, app-only ordering, supermarkets that feel more like logistics hubs than public spaces. The dystopia doesn’t arrive; it updates.

The social side-effect is just as grim. If convenience formats dominate everyday life, “not having to talk to people” can mutate into “barely being able to talk to people”—because the public spaces that train everyday social interaction shrink. This fear isn’t pulled from nowhere: many societies already wrestle with loneliness and social isolation, and some cultures have built entire convenience ecosystems around solo routines. The trend of solo dining is global and accelerating; even the backlash against solo diners tells you how structural this shift is.
Japan’s policy discourse also treats social isolation as a serious issue, with estimates in the seven figures depending on definitions and age ranges.

Best case: the same technologies do something very different.
AI becomes backstage infrastructure, not the star of the show: reducing waste, smoothing prep, improving forecasting, preventing burnout. Automation takes the dull, repetitive, error-prone parts; people get more time for what guests actually remember—attention, recovery when something goes wrong, and craft. “Augmentation” becomes the guiding principle: fewer people can do better work, faster and more consistently, because the system supports them. And crucially, that regained efficiency can be used to re-open the middle: not by making everything cheap, but by making warmth and quality less exclusive again.

Here’s the crucial difference: efficiency gains are reinvested, not extracted. Better planning means less chaos, more stable shifts, more training, and less turnover. Guests still get speed where they want speed—but also warmth where warmth matters. This isn’t naïve: even giants that tried to “tech their way out” of labour intensity have had to admit that hospitality is not only a workflow problem. Starbucks publicly reversed aspects of its labour-cutting/automation emphasis and leaned back toward staffing and in-store experience, explicitly framing technology as support rather than replacement.

So what is the likely middle path?
Reality will probably become hybrid. We’ll see more self-service ordering, more convenience formats, and more automation—especially where speed is the core job to be done. At the same time, we’ll see a stronger hunger for places that feel human. Paradoxically, technology may increase the value of “real human attention,” because once it becomes rarer, it becomes more noticeable. The risk is not that automation exists; it’s that it becomes the only affordable lane, while everything “human” silently migrates into premium positioning.

The outcome depends on choices we still control:
Do we automate transactions or relationships? Replacing payment friction is one thing; replacing care is another.
Do efficiency gains become extraction—or reinvestment? The same forecasting tool can either stabilize schedules or justify running permanently understaffed.
Do we design for trust and transparency—or for surveillance by default? People tolerate data collection differently when the benefit is clear, optional, and minimal.
Do we treat the “middle” of hospitality as disposable—or as a civic space worth protecting? A society without affordable third places becomes colder even if it becomes more “efficient.”

If we want a future where warmth remains affordable, we need to treat digital transformation as a tool for inclusion, not a shortcut to a two-tier society. The future of dining isn’t a prediction. It’s a design problem—economic, ethical, and deeply human.

Jonas Legelli

Author’s note / use of AI tools

All ideas, observations, and opinions in this blog post are my own. I used AI (ChatGPT 5.2) as a support tool for language polishing and for generating imagery. 

References & further reading

Leave a comment