How we built Tišina

Tišina means "silence" in Croatian. Before we wrote a line of code, that word was the whole brief.

Most restaurant sites solve a narrower problem than they think. They show you hours, a gallery, a PDF menu if you're lucky, and a button that calls the restaurant. That's a fine site for a place that wants you to already know what you're walking into. It's the wrong site for a concept built around pace — where the point is that you're not in a hurry, and neither is the kitchen.

The brief we gave ourselves

Tišina is one of our own concepts: a slow-dining restaurant built around regional ingredients and a dining room that seats 28 people, on purpose. Small enough that every table matters, small enough that the booking flow couldn't just be a generic reservation widget bolted onto a template.

The challenge wasn't technical in the usual sense. It was that "quiet" and "patience" are feelings, not features, and a website has about four seconds to communicate a feeling before someone decides whether to keep scrolling.

What we actually built

The structure is simple on purpose. One scroll, four movements: the room, the food, the people behind it, the table. No navigation maze, because a restaurant website doesn't need an information architecture — it needs a reason to stay on the page for thirty more seconds.

The transitions carry the tone. We built the site on Next.js with Framer Motion specifically because CSS transitions alone couldn't give us the kind of cross-fade pacing we wanted between sections. Most scroll reveals use a linear or ease-out curve because it's the default — we didn't. We hand-tuned a custom cubic-bézier that front-loads the motion and lets it settle slowly, the same shape you'd use for a camera rack-focus, not a UI toast. Slow enough to feel unhurried, fast enough that nobody thinks the page is broken. That distinction sounds small. On a slow-dining site, it's the whole point: if the motion feels rushed, the restaurant feels rushed, and the concept falls apart before anyone reads a word of copy.

The menu is typography, not a PDF. We built an interactive menu focused entirely on type and spacing — no food photography competing for attention, no icons. Seasonal ingredients and regional heritage get to be the content instead of decoration around the content.

Booking is short because the room is small. With 28 seats, there's no reason to make someone click through five screens to request a table. The booking flow is built around that fixed number — not a generic third-party widget that doesn't know or care how big the room actually is.

The decision that mattered most

We put the images on Vercel's edge network deliberately, not as an afterthought. Food photography is the heaviest thing on a restaurant site, and the people most likely to visit — scrolling on their phone, deciding where to eat tonight — are the ones with the least patience for a slow load. Largest Contentful Paint is the metric that actually captures this: it measures when the biggest, most meaningful thing on the screen finishes rendering, and on a restaurant site that's almost always the hero photo, not the headline. A beautiful photo that takes three seconds to appear has already lost — both the visitor and the score. Serving pre-sized WebP from the edge, close to wherever the visitor actually is, is a boring decision. It's also the one most restaurant sites skip, which is exactly why it's the one that matters.

What I'd tell someone building something similar

If your product is a feeling before it's a feature list, don't start with the page structure. Start with the word that describes the feeling, and let every decision — the transition timing, the font weight, how much white space surrounds the menu — answer to that word. "Fast and modern" is what everyone's brief says. It wasn't ours, and the site is better for it.

Tišina is a concept build — not a live client restaurant — which means we got to make every one of these calls without a stakeholder asking for a louder hero image. That's the advantage of building something to show what's possible rather than retrofitting someone else's existing brand. The next version of this, for a real kitchen, starts from everything we learned here.