Built with
// Case Study
An AI chatbot recommending underground electronic-music events scraped from four major ticketing platforms, with genre classification, DJ lineup extraction, and vector similarity search in one platform.
The problem
No single tool existed for discovering underground EDM events across all the major ticketing platforms. Each platform had a different data structure, none shared an API, and DJ lineup data was not standardized between them.
A taste-based recommendation engine meant solving scraping, classification, and search all at once, then surfacing the result in a way a fan could actually talk to.
A Python and n8n pipeline pulling events from four platforms through platform-specific APIs, normalising four different data structures into one consistent event schema.
Each event is classified by genre, and DJ lineups are extracted and normalised across sources so the same artist reads the same way no matter where the event was listed.
Events are stored in Supabase with vector embeddings, enabling taste-based similarity search rather than simple keyword matching, so a fan gets events that feel right.
A React frontend with an OpenAI-powered chatbot surfaces relevant events from natural-language taste descriptions, turning a plain-language request into a curated set of nights out.
Four major EDM ticketing platforms unified into one searchable database.
DJ lineups extracted and genre-classified automatically across every source.
Vector similarity search enabling taste-based discovery instead of keyword lookup.
An AI chatbot handling recommendations in natural conversation.
Built with
B2B / AI Services / US
A LangChain agent on Telegram qualifying B2B leads across 15 AI service types and booking meetings with Meet links.
Healthcare / France
A custom French and English e-learning platform with signed-token video protection and a GHL patient journey.
Home Services / UK
A single AI agent and knowledge base deployed across WhatsApp, SMS, website chat, and Messenger.
// Free Audit
Book the audit. Thirty minutes. Free.