Theatro
Telegram chatbot that answers questions about TV and entertainment using a Neo4j knowledge graph.
Theatro was built as an early AI chatbot for entertainment discovery, translating user questions into graph-backed queries and returning structured answers through conversation.
The stack used Dialogflow for intent handling, GraphQL for the query layer, and Neo4j as the graph database behind the knowledge space.
Conversational model
The objective was to go beyond scripted TV bots and build something closer to an AI entertainment assistant, where users could ask naturally about shows, actors, genres, or relationships and still get structured answers back.
That meant connecting conversational input with a graph-aware data layer so the chatbot could answer with more context and flexibility than a keyword-driven discovery flow.
Architecture
Users interacted through Telegram, which connected to Dialogflow for intent analysis. From there, the webhook layer translated requests into GraphQL and Cypher queries, then pulled the final answer from Neo4j.
The graph stored relationships between shows, actors, and genres for the chatbot to query.
Selected material
Get in touch.
Based in Barcelona. If you want to talk performance, memory systems, or something I have built, my inbox is open.