From conversational interfaces to serverless natural language APIs

on YouTube ↗32m

" summary

Built a natural-language-first app called TravelLevel with web, SMS (Twilio) and Alexa interfaces that reliably records travel dates. Intent detection uses two intents handled by simple regular expressions; entity parsing uses a custom open-source date parser called roller because Dialogflow/LUIS/Watson missed multi-day and contextual dates. Backend is fully serverless on AWS: CloudFront + S3 static React, API Gateway + Lambda, Cognito, DynamoDB and IoT/WebSocket notifications. It works in production, remains in use, and costs under $1/month for current traffic.

AI-written summary · about

" chapters (5)

1. 0:00Start
2. 18:26web server
3. 21:15authentication
4. 25:39websockets
5. 28:01deployment

" related talks

• 33mYour e-commerce botFabrizio Ciacchi · code.talks commerce 2018
• 01h 00mYour next app might be a bot! Building conversational UX with API.AIDan Aharon · Google Cloud Next '18
• 19mGo from idea to AI-powered app in minutes | AWS Developer Day 2025Hazel Saenz · AWS Developer Day 2025
• 50mSpeak To Me: Voice App Conversation PracticesHeather Downing · NDC London 2018
• 54melevating customer experience w/ conversational AIArjun Chatterjee · AWS re:Invent 2019
• 52mThe most natural UI. Building voice-enabled experiences with Amazon AlexaAndrea Muttoni · code.talks 2017
• 34mOk Google! We need to talkEliza Camber · GDG DevFest Ukraine 2017
• 46mWELCOME TO THE AGE OF CONVERSATIONAL INTERFACES!Rabeb Othmani · NDC London 2018