I Had This Idea at MIT 10 Years Ago. AI Just Let Me Finally Build It (Without Going Broke or Crazy).

Richard de Vries — Tue, 20 Oct 2026 01:02:38 +0000

Back in 2016, during a course at MIT, I had an idea born entirely out of pure, unadulterated menu anxiety.

I was sitting at a restaurant, staring blankly at a massive menu, realizing a universal truth: I seriously never know what to order.
You look up a spot on Yelp or Google Maps, see a glowing 4.5-star aggregate, drive over, and end up ordering a completely mediocre entrée. Meanwhile, a dingy neighborhood dive with a 3.8-star score might quietly be serving the single best charred cabbage or slice of pizza in the city.

The atomic unit of food discovery is broken. You don't eat a restaurant… you eat a dish.

Back then, as a designer, the idea felt too complex to build by myself. A map layer, user-generated reviews, structured data, maps APIs, auth, state management… it required a real engineering team. So, like a thousand other design concepts, it lived in my head rent free.

Fast forward a decade. I'm now a Senior UX Designer and Chapter Lead for Digital Design at Philips. I've built side projects before: years ago, I experimented with lightning-fast rapid prototyping through a project called 400Minutes, where I came up with a super simple product idea and launched it the same week (in 6.6 hours: 400 minutes)

About a year ago, I decided to stop prototyping in Figma and actually build Chef's Special myself using modern stack tools (React, Vite, Supabase, and leaning on Claude for the code architecture).
And boy, did reality hit immediately.

Pivot #1: Escaping the Two-Sided Marketplace Trap

As a UX designer, you look at cold-start problems analytically. My initial instinct was simple: Build a clean map of restaurants, slap some review fields on it, and users will flood in.
Spoiler alert: They didn't.

Building a two-sided marketplace (writers of reviews vs. consumers of reviews) when you are starting from scratch is a classic designer delusion. If nobody is writing reviews, the map is empty. If the map is empty, nobody visits.

So over the past year, I pivoted. I realized that forcing users to write restaurant reviews is a chore. But passionate foodies love documenting what they eat and what they cook.

The Single-Player Utility: I expanded the app beyond just restaurant maps. I added an Instagram-style visual feed and a way to log recipes cooked at home. Suddenly, the app became useful even if you were the only person using it-serving as a personal culinary journal and shareable food map.

The "Chef Specialist" Crown: To make engagement fun without relying on boring text reviews, I introduced a Foursquare-style crown mechanic. When you discover and rate a standout local plate, you can claim and defend the crown for that specific dish.

Personal Food Map: After I had used this for a while, I noticed it was actually kind of fun to watch where have eaten before. So I added food maps on the profile page, look at mine for example: https://chefsspecial.app/profile/richard

The Designer-as-Maker Reality Check

Making the jump from designing a product to shipping and maintaining it for a full year teaches you things Figma never will:
You design fewer, better things: When you have to write the code yourself, feature bloat dies a fast death. Every extra button you add is 30 minutes of debugging state management.

Designers make terrible scope managers: I spent way too many weekends tweaking UI micro-interactions when I should have been fixing edge cases in geospatial map queries. Old habits die hard.

Expectations management: Am I quitting my day job to make a million dollars off Chef's Special? Absolutely not. This isn't a venture-backed hyper-growth play. It's just me having fun over the last year, scratching an itch I've had for a decade, and treating code as a medium for design.

What I Learned

If you're a designer sitting on an idea you thought was "too complex" a decade ago, the barrier has never been lower. You don't need a massive engineering budget to test your hypotheses anymore; you just need curiosity, stubbornness, and a deep hatred of ordering the wrong thing off a menu.

If you want to check out the chaos, explore the Twin Cities dish map, or see how a designer's side project evolved over 12 months of building, take a look at chefsspecial.app.
And drop a comment below: What's an idea you sketched out ten years ago that you're still waiting to build?

DEV Community: Richard de Vries

I Had This Idea at MIT 10 Years Ago. AI Just Let Me Finally Build It (Without Going Broke or Crazy).

Pivot #1: Escaping the Two-Sided Marketplace Trap

The Designer-as-Maker Reality Check

What I Learned