From a simple chatbot to a fully programmable dating coach
summary
A virtual coach called Lara evolves from a Messenger prototype into a multilingual, embeddable dating coach: initial build uses Dialogflow + middleware connected to Meetic APIs and yields a 30% higher registration rate from a conversational form. Team adds custom NLP to extract intents and entities (e.g. “blue eyes”, height > 180 cm), small talk, profile optimization and icebreakers, then internalizes tooling, creates a conversational designer role and a JavaScript SDK for iOS/Android/web. System sends about 2 million messages a day and is reused by Japanese teams with local NLP.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 1:03From simple chatbot to Lara
- 2:46Conversational form boosted registrations
- 3:28NLP for profile searches
- 5:54Small talk and humor in chatbot
- 6:56Role of conversational designer
- 7:59Lara on Google Home
- 10:24Scaling Lara to full coach
- 11:51SDK for multi-platform coach
- 13:40Globalizing Lara in six languages
- 14:01Japanese Lara launch
- 15:25Summary and impact
- 16:05Q&A on personality matching
- 17:47User feedback for new features
- 20:59Q&A on negative behavior
related talks
22mOvercoming Loneliness with LLM Dating Assistant
50mThe Art of Coding a Conversation: Designing Bots
27mA primer to write a conversational chatbot
18mChatbots and the future of online learning
33mYour e-commerce bot
24mConversational Interfaces for Chatbots and Artificial Intelligence
46mChatbots, shall we shift into second gear?
34mAdopt an AI driven chatbot today