🏢 Cafe / Restaurant · Reservoir, Melbourne
Reservoir · Local Business
Research Phase
Before writing a single line of code, our research agent studied Alour Cafe — their current website, social presence, reviews, team, and services. Here's what stood out.
4.3 stars from 398 Google reviews, plus a 4.2 rating from 769 evaluations on lacarte.menu — making it one of the most-reviewed cafes in Reservoir with over 1,100 combined ratings across platforms
Owner Peter is known by name in reviews for greeting regulars personally — operating from 251-255 Broadway with 6am-5pm hours 7 days a week, including outdoor seating for dine-in and takeaway
Menu spans breakfast through lunch with items like crispy chicken, omelettes, waffles, tostadas, and mushroom dishes — yet the cafe has no website at all, relying entirely on walk-in traffic and platform listings
Design Phase
Every design decision was intentional — informed by the research, not pulled from a template library.
Visual Direction
A 6-page cafe site with 9 components across home, about, menu, gallery, functions, and contact pages. The bespoke CoffeeQuiz interactive component matches visitors with their ideal coffee order — a playful engagement tool that reflects the cafe's friendly, personal atmosphere. The MenuSection component structures their diverse food and drink offerings.
Design Rationale
Alour Cafe has 1,100+ combined reviews across Google, lacarte.menu, Yelp, and TripAdvisor but zero web presence — all that social proof is scattered across platforms they don't control. This design consolidates their identity into an owned channel, with the CoffeeQuiz capturing the warm, personal touch that reviewers associate with owner Peter. The functions page opens a new revenue channel for a venue with outdoor seating and capacity for private events.
Build Phase
Site Architecture
// Next.js App Router
Methodology
Spark (research agent) scraped Alour Cafe's current website, Google reviews, social media, and business listings. Every piece of real data — team names, service pricing, business hours, unique details — was extracted into a structured profile.
Based on the research, a bespoke design direction was created specifically for Alour Cafe. No templates — the typography, color palette, layout patterns, and visual effects were all chosen to match this business's personality and industry.
Pixel (design agent) built the entire 6-page Next.js application from scratch — 9 custom components, real content from the research phase, App Router with proper routing, and production-grade code.
Sentinel (QA agent) verified the build compiles cleanly, checked for placeholder content, validated all routes render correctly, and confirmed the design is unique (not recycled from another demo).
Additional Notes
Research drew from 12 sources — the most comprehensive in this batch — including Google Maps (4.3 stars, 398 reviews), lacarte.menu (4.2 rating, 769 evaluations, menu categories), Yelp (owner name 'Peter' from review snippet, 4-star rating), TripAdvisor via Yahoo Search (hours confirmation, 4/5 rating), and Cylex directory listing. Data completeness rated medium despite no website existing. The Brave Search API was exhausted, so discovery used Yahoo Search and web_fetch fallbacks. Owner name and personal greeting style were confirmed from Yelp review content.
More from the experiment
This case study is part of the AI Web Lab — a 30-day experiment where AI agents build bespoke websites for real Melbourne businesses. Every site is researched, designed, and built from scratch by a team of five specialist AI agents. The full methodology, code, and results are documented transparently.
This study is being prepared for publication as part of a research paper on AI-assisted web development at scale.
We built Alour Cafe's site in 10 minutes. Same process, same quality — a custom website built for you, not a template.