NaijaNest AI

// stack

// architecture

// overview

A Node.js, Express and MongoDB backend that feeds real listing and neighbourhood data into GPT-4o to produce locally-grounded market analysis for Lagos, Abuja and beyond. The design problem was hallucination: a property AI that invents prices is worse than useless in a market with little public pricing data. The answer was a dynamic market context builder that continually aggregates current rent averages, fees and yields from the database into the prompt, so every verdict is anchored to real figures. On top of that sit listing price evaluation, deep property analysis including the upfront costs Nigerian renters actually face (agency, legal, caution, service charge), conversational natural-language search with constraint relaxation when nothing matches exactly, and honest neighbourhood summaries.

// what was built

• ·Dynamic market context builder aggregates live rent averages, fees and yields from the database into every prompt, so the model reasons over real figures rather than inventing them.
• ·Listing price insight evaluates a proposed price in real time and returns a verdict — competitive, overpriced — with market comparisons and actionable guidance.
• ·Deep property analysis covering investment potential, lifestyle match, pricing verdict, area highlights and the hidden upfront costs specific to the Nigerian market.
• ·Smart natural-language search extracts intent from conversational queries and relaxes constraints gracefully when an exact match does not exist, instead of returning nothing.
• ·Neighbourhood intelligence models storing average rents by property type, safety scores, commute times and yields, powering localised summaries.
• ·Full CRUD for sales, rentals and shortlets, with JWT authentication.