I don't need friends, I can build my own
summary
Builds a retrieval‑based travel recommender in Python using tokenizers, nltk and scikit‑learn prototypes then migrates to rasa for intents, entities, slots and a holiday form, adds a fallback that triggers under 40% certainty, and deploys with Rasa X (Docker + GitHub) to collect real conversations. Claims: memory via slots and forms improves recommendations; reality: it recommends reliably on scripted paths, fails off‑path and needs human‑trained examples to improve.
AI-written summary · about
chapters (14)
- 0:00Start
- 0:10Sarah Levins
- 0:53Chat Bots
- 4:49Chat Bots Have a Front End and a Back End
- 5:47Intent
- 6:25Entities
- 7:24Python
- 10:32Actions Py
- 11:24Conversation Design
- 12:33Training Data in the Nlu
- 13:46Custom Action
- 18:47Out of Scope
- 19:04Summarize
- 23:59Github Actions
related talks
01h 31mGoing beyond “Sorry, I didn’t get that” building AI assistants that scale...
01h 14mBuilding a Recommendation System with a Hybrid Search From Scratch
35mNatural and conversational shopping experiences
44mThe RAG Really Ties the App Together
28mWhat the RAG? Demystifying Retrieval Augemented Generation
57mCrafting Intelligent Agents with Context Engineering
01h 23mBuilding an AI Travel Agent That Never Hallucinates
48mUnlocking LLM Potential: From Exploration to Integration