AI-Powered Event Discovery for Buenos Aires
Turned fragmented Buenos Aires event data into a live discovery product that helps people find plans through grounded AI search, not another dead-end chat demo.
Outcome
Launched a working event discovery product with 800+ events and 190+ venues pulled into one searchable experience instead of forcing users to jump between sources.Best fit
Teams building data-heavy discovery products that need grounded answers and usable product workflows.Company
Personal ProjectSummary
Buenos Aires event information is scattered across Instagram stories, Facebook events, Eventbrite, Passline, venue sites, and WhatsApp groups.
I built BA Eventos as a live discovery product that turns fragmented local data into a searchable experience with grounded AI answers.
About
The city has a dense cultural calendar, but the information lives across incompatible sources and changes constantly.
People do not search for events like database records. They ask for moods, neighborhoods, time windows, and social context.
Business Objective
The product needed to help people find real plans without forcing them to jump between social platforms, ticketing sites, and venue pages.
That required a data pipeline, an admin surface, and a retrieval layer that could answer natural questions without hallucinating dead-end recommendations.
Product Direction
Treat event discovery as a data-quality problem first.
The product could only feel useful if the underlying event and venue data was current, matched, and clean enough to search.
Use AI where intent matters.
Semantic search handled queries around vibe, neighborhood, and timing, while structured filters kept the answers grounded in real records.
Product Surface
The product is a working example of AI-assisted discovery over messy local data, not just a chat interface.
The Work
Turned scattered sources into a usable event catalog.
The system ingests events, matches venues, enriches descriptions, and keeps the catalog organized enough to support real search.
Built discovery around plain-language intent.
Users can ask questions like romantic jazz in Palermo or underground techno this weekend while the product stays tied to real event records.
What Shipped
BA Eventos shipped as a live example of an AI product where the hard part is the operating system around the model: data ingestion, cleanup, retrieval, and a usable product surface.
Data coverage
Hundreds of events moved into one searchable product.
The product launched with 800+ events and 190+ venues organized into a single discovery surface.
Grounded answers
AI search stayed tied to real event data.
Semantic search and structured filters helped answer intent-heavy questions without drifting into unsupported recommendations.
Admin tooling
The catalog could be kept usable.
Admin and data-quality tooling supported venue matching, enrichment, and cleanup behind the public experience.
Reusable pattern
The build proved an AI discovery pattern.
Ingest, clean, enrich, embed, then answer with grounded results instead of shipping another empty chat box.
Product Screens
The product surface shows the grounded discovery experience built on top of the event and venue catalog.