A plain-language tour of the app — where it started, what it does today, the decisions behind it, and the road ahead. No jargon required.
The Brick Drop tells you how a LEGO set you're into actually rates — one honest letter grade built from real reviews, colored by the set's own photo, with a dry-but-hyped take on whether it's worth it. It's live on the web, every set has its own shareable page, and it installs like an app.
Think of it as a movie-rating site, but for LEGO sets — and the score isn't one critic's opinion, it's distilled from a whole pile of real reviews.
Every set gets a single Drop Score (a letter grade like A− or C+). Behind that letter, the app mines LEGO reviews, reads the sentiment, and rolls it up into the grade. Alongside it, three quick "reads" answer the questions people actually ask: Is the price fair? · Will you love it? · Holds its value?
Two things make it feel like The Brick Drop and not a spreadsheet: the grade letter is colored by a shade pulled from the set's own photo, and each set carries a hot-take written in the brand's group-chat voice — opinionated, honest (it'll call a mid set mid), never corporate.
The old tool grew by accretion — a calculator, then an app, then a discovery layer, then a brand revamp piled on top. It got noisy and the grade got buried. This is the coherent reset: open to the grade, let the photo and the letter do the talking, be honest about what we don't know.
The app is a real, installable Next.js web app. You can search a set by name or number, browse a grade-sorted grid, and land on a styled result page with the grade, the reads, and the take. Share any link and it unfurls as a branded card.
Over one long build we took the card from a bare skeleton to a polished, personality-filled product. Grouped by what it did for you:
| Slice | What changed | Why it matters |
|---|---|---|
| Hot-takes | Every set now has an AI-written take + three read blurbs + per-lens "reasons," in your Instagram voice | The card had empty text lines — now it has personality and a point of view |
| Auto colors | Every graded set's letter is colored by a vibrant shade sampled from its photo (your 6 hand-picks still win) | Grades used to be uniform yellow; now Pikachu is Pikachu-yellow, Tintin is rocket-red |
| Real framing | Product photos show the whole set on white (no more weird cover-crops); your curated crops still apply | Wide sets like the Helicarrier were getting sliced in half |
| Slice | What changed | Why it matters |
|---|---|---|
| Phosphor icons | Swapped the read-row emoji for clean Phosphor icons (incl. the LEGO minifig-head for "Will you love it?") | Emoji off-brand; icons look intentional and are on-theme |
| Grade-sorted Browse | The grid leads with the best grades instead of A–Z | The strongest sets surface first — a far more useful shelf |
| No price-per-piece | Banned "cents a piece" everywhere and regenerated any take that slipped | It contradicts your value philosophy — value isn't per-brick math |
| Slice | What changed | Why it matters |
|---|---|---|
| Link previews | Every shared /set link unfurls as a branded grade card (real Avenir Next Condensed, not the raw photo) | Passive reach — every paste in a group chat looks like The Brick Drop |
| 4:5 share card | A "Download share card" button saves the image-forward Instagram card as a PNG | Your north-star DNA — a polished asset ready to post |
| Layering refactor | Cleaned up an architecture wrinkle so the data layer no longer depends on business logic | Unblocks the next big feature: personalization |
Nine of these shipped in a single session, each as its own reviewed pull request. The whole thing stayed green — 509 tests, clean type-checks, clean production build — the entire way.
Data flows one direction: from a database of pre-computed grades, through a translation layer, into the pages you see.
The grading engine is the kitchen — slow, careful, done in back. The database is the pass where finished plates wait. The view model is the server who carries the exact same plate to every table. You (the diner) never see the kitchen chaos — just a consistent dish, fast.
These aren't preferences — they're now enforced in code and saved so they never regress:
What's deliberately not done yet — stated plainly so nothing hides:
Personalization isn't reachable yet. The engine can grade "for a display collector" vs "for a builder," and the per-lens reasons are already written and stored — but there's no onboarding to pick your lens, and only the general grade is stored per set. So "A− for you · B+ overall" exists in code but nobody can see it. This is the next big unlock.
Only 28 sets are graded. Browse is small until we grade more of the catalog.
The 4:5 card uses generic framing. The image engine can't apply your Crop Curator zoom/position, so wide sets crop. A "come back later" item.
No predicted grade for unreleased sets and True Value is still just a one-line read, not a full price breakdown.
The layering cleanup just unblocked the feature the whole app is named for: how a set rates for you.
Start the personalization arc — per-lens grades first, then onboarding. It's the last big unbuilt piece, it's finally unblocked, and it delivers the promise the name makes.
The Brick Drop · built with receipts, not vibes