Homiq

The property market faces a significant challenge: reliable and up-to-date information about homes is often fragmented, inconsistent, and difficult to access. Renting or buying a property becomes complicated when key details such as nearby schools, safety, or weather conditions are not easily available or regularly updated. Our project addresses this issue by creating an interactive web application that helps users find homes based on their specific needs—such as being close to good schools, having access to gyms, or enjoying pleasant weather—while also answering follow-up questions about each property or its surrounding area. The system uses Perplexity’s Sonar Pro API to perform real-time research and generate clear, evidence-based responses, ensuring that the information shown is accurate, current, and trustworthy.

We exploit Perplexity’s strongest features—its retrieval-augmented reasoning and structured output capabilities. The API’s live web search ensures that all property data and contextual information are always up-to-date, even in a rapidly changing market. Its reasoning capabilities allow the system to interpret complex user queries, combine multiple data points, and deliver concise, relevant answers supported by citations. By structuring the API responses through JSON schemas, we can automatically organise the information into consistent, easy-to-display results that adapt to different user questions. Perplexity’s ability to retrieve and reason over current information allows our application to bridge the gap between static property listings and real-time, contextual understanding, transforming property searching into an intelligent, conversational experience that delivers clarity, trust, and interactivity to users.

The web application itself is built as a TypeScript React front end that composes a zoomable, draggable canvas of informational widgets generated from AI-driven research. A global state manages widget collections, active selections, and chat history, while a responsive grid arranges results dynamically in a bento-style layout. The interface merges predefined widget templates with live API data—titles, markdown summaries, and citations—and supports conversational input for searches and follow-ups. This modular, accessible design makes the system flexible, visually clear, and easily extensible for future data sources.

Built With

• perplexity
• typescript