No. 03
AI & UX
Vindmolen
Vindmolen is an AI-prototyped concept, prompted to improve the rental search experience. Vindmolen acts as an aggregator, providing relevant leads and AI-assisted applications to help ease search fatigue.
Role
UX Designer
Timeline
3-week sprint
Team
Solo project
Year
2026
The problem & the user
The Challenge
The Dutch rental market has devolved into a high-stakes arms race where speed is the only metric for success. Renters are forced to juggle multiple subscriptions and are trapped in a relentless 24/7 loop of alerts. This email roulette creates a bottleneck: with viewings often fully booked within minutes, even the most diligent seekers can spend months without a single invitation to visit a property.
The User
I designed for the high-intent searcher, typically a digitally native young professional or expat. They are exhausted by sending a thousand love letters to landlords, only to be met with silence. They need to know if a property matches their specific criteria before investing energy into an application.
In numbers
400k
housing deficit, creating a climate of scarcity
1/40
chance of viewing per application
150+
instant inquiries for an Amsterdam apartment
The approach & process
01
Research & Discovery
When formal API access from the major providers was unavailable, I developed a web scraping strategy focused on public metadata. For this approach, I needed to ensure the project remained non-disruptive and legally sound under EU Database Rights. The final concept functions as a referral layer, extracting only the necessary data to guide users back to the original listings.
02
Tooling & software
I utilised Cursor Pro to bridge the gap between prototype and functional code without needing deep technical knowledge. By using Composer, I created an agentic workflow that translated my intent into complex feature implementations. This approach, powered by Claude 3.5 and GPT-4o, ensured context-aware development throughout the entire project.
03
Prototype & Build
I built Vindmolen using Cursor to bridge Next.js and FastAPI, creating a rental platform that feels both “Dutch-rooted” and “tech-forward.” Behind the scenes, secure Python agents handle the heavy lifting of scraping and geocoding. This resulted in a functional platform that acts as a lead funnel for landlords, and an attractive all-in-one solution for those searching for an apartment.
04
Brand & UI system
The name Vindmolen combines vind (find) and windmolen (windmill), a Dutch-forward name for a Dutch rental problem. The logo is built around a negative-space X that reads three ways at once: Amsterdam’s city cross, “X marks the spot” for discovery, and the crossed blades of a windmill. House silhouettes frame the symbol so it stays anchored in housing, not abstract tech. Rather than another sterile, blue property site, the interface borrows inspiration from premium fintech dashboards: JetBrains Mono for crisp, data-dense type, and a loud-but-controlled palette of lavender, lime, and neon pink so listings and filters stay scannable under stress. The aim is simple: speed up comprehension and give renters a tool that feels focused and professional.
The solution
Vindmolen is a rental search platform designed to turn the friction of the Dutch housing market into a structured, AI-assisted workflow.
Onboarding
A brief, guided introduction that establishes the product’s purpose. These optional steps allow users to define their search criteria immediately, ensuring maximum value from day one.
Listings
Designed for rapid orientation. High-impact listing cards prioritise essential details, allowing users to scroll, click, and apply without wasting time.
Search Filters
An interactive map-based interface focusing on Amsterdam, The Hague, Utrecht, and Rotterdam. Users can filter by neighbourhood, price, and size, with support for multiple search profiles.
Renter Profile
A structured, multi-step journey that tracks profile completeness. This vital information is the data source that powers the platform’s AI.
Deep-Match
Advanced AI letter generation. It performs a multi-point handshake between the Renter Profile and specific Listing Metadata to create highly tailored applications.
Quick Fill
A high-speed alternative to Deep-Match. This structured template uses dynamic fields (like [STREET] or [RENT]) that are autofilled with the listing’s metadata.
Tracker
The Tracker logs applications and favourites, and uses your activity to predict your chances of securing a home. On the backend, the same data personalises “For You” suggestions.
Onboarding
A brief, guided introduction that establishes the product’s purpose. These optional steps allow users to define their search criteria immediately, ensuring maximum value from day one.
Listings
Designed for rapid orientation. High-impact listing cards prioritise essential details, allowing users to scroll, click, and apply without wasting time.
Search Filters
An interactive map-based interface focusing on Amsterdam, The Hague, Utrecht, and Rotterdam. Users can filter by neighbourhood, price, and size, with support for multiple search profiles.
Renter Profile
A structured, multi-step journey that tracks profile completeness. This vital information is the data source that powers the platform’s AI.
Deep-Match
Advanced AI letter generation. It performs a multi-point handshake between the Renter Profile and specific Listing Metadata to create highly tailored applications.
Quick Fill
A high-speed alternative to Deep-Match. This structured template uses dynamic fields (like [STREET] or [RENT]) that are autofilled with the listing’s metadata.
Tracker
The Tracker logs applications and favourites, and uses your activity to predict your chances of securing a home. On the backend, the same data personalises “For You” suggestions.
Reflection & learnings
What I learned
Building Vindmolen was my first solo 0-1 project, instilling in me the power of ownership and the necessity of technical adaptation. By utilising Cursor Pro, I have learnt to manage agentic workflows that bridge the gap between product intent and functional code. Navigating EU Database Rights independently also sharpened my ability to solve complex legal constraints through creative technical solutions.
Changes I’d make
In future iterations, I would partner with an engineer to apply a four-eye principle to backend security and architectural decisions. While the prototype is functional, professional validation would ensure the system is built for scale, particularly for the automated notification system & web scraper etiquette. I would also allocate dedicated time for formal user testing rounds to refine the UX/UI and expand the brand strategy into a comprehensive marketing plan for a public launch.