GreyStone

Inspiration

Increasingly, real estate decisions are being made with area level structural factors, not just individual property deals. Capital discipline is becoming more and more critical to avoid areas with poor council regulation, weak demand, and other socioeconomic factors. Public data on various categories (demographics, rents, planning and transactions etc.) are de-centralized and fragmented. Therefore, utilizing the proprietary datasets provided by ibEX has enabled us to produce an all in one platform for identifying and deciding favorable locations to invest in.

What it does

The project allows users to select any area globally, assessing its attractiveness for real estate investment. It analyses various factors: demographic, rental, supply, transaction and regulatory signals, to produce an informed outlook and score on a property's attractiveness. The raw data is normalized, and then scored across different dimensions: demand strength, supply constraint, liquidity and policy risk. Finally, its converted to a human friendly form for easier interpretation.

How we built it

We built Use‑Change Hunter as a full‑stack geospatial intelligence platform that analyses UK planning activity to surface high‑probability property value‑add opportunities. The frontend was developed using Next.js (React + TypeScript) with Leaflet for interactive map visualisation, enabling users to search by postcode or location and explore nearby planning signals. The backend was implemented in Python using FastAPI, where we created a modular service layer to integrate the Ibex API for live planning application data. To manage limited API credits and improve performance, we implemented a Redis caching layer alongside a PostgreSQL/PostGIS database for persistent storage and spatial querying. A custom rule‑based scoring engine processes nearby approved and refused precedents, keyword‑matches proposal text, and weights recency and density to generate opportunity, risk, and confidence scores. We also built a transparent scenario modelling module that estimates GDV and rent uplift based on configurable assumptions. The entire system was containerised with Docker for reproducible deployment and designed with a modular architecture so additional datasets (e.g., EPC or price paid) can be integrated in future iterations.

Challenges we ran into

Since we had limited API credits, we had to design a caching layer (Redis + DB) to avoid repeated calls and optimise queries. Balancing real-time data vs cached results was a key challenge. We also had to prioritise core features (planning-based scoring + map + scenario model) and cut or simplify others (full policy integration, perfect valuation models). Finally, some planning and property datasets are incomplete or inconsistent across areas, so we had to design the system to work with partial data and still provide useful outputs.

Accomplishments that we're proud of

We’re proud of turning complex UK planning data into a clear, actionable product that highlights real property opportunities rather than just displaying raw information. Within a short timeframe, we built a complete end-to-end system, integrating the Ibex API, developing a scoring engine, and delivering an interactive map-based interface.

A key achievement was creating an explainable opportunity scoring model based on planning precedents and approval trends, making the tool both practical and trustworthy. We also handled technical constraints effectively by implementing caching to optimise API usage.

Finally, by adding scenario modelling, we went beyond analysis to help users understand potential financial outcomes, making the product genuinely useful for real-world decision-making.

What we learned

Through building Use‑Change Hunter, we learned how complex and fragmented real-world planning data can be, and the importance of designing systems that can interpret messy, unstructured inputs. We gained hands-on experience integrating external APIs at scale, particularly the need for efficient caching and thoughtful query design when working with limited credits. The project also deepened our understanding of geospatial analysis using PostGIS and how spatial context drives meaningful insights in property intelligence. Perhaps most importantly, we learned the value of explainability — ensuring that every score is transparent and defensible — and how to balance technical ambition with pragmatic scope under tight hackathon time constraints.

What's next for GreyStone

Looking ahead, we plan to evolve Greystone from a hackathon prototype into a production‑ready intelligence platform for property developers and investors. Our next steps include expanding geographic coverage, integrating richer datasets such as EPC and sold‑price data, and refining our planning probability models using historical outcomes. We also aim to enhance the user experience with portfolio‑level scanning, automated lead generation, and more advanced financial modelling. Ultimately, our goal is to position Greystone as the go‑to decision-support tool for identifying high‑potential property opportunities before they appear on the wider market.