Open AI Assistants
summary
Claims a person can create your own assistant in 25 minutes and test it five minutes later using the open source SUSI.AI stack; implemented components include a search-aggregation engine, JSON-path access to external APIs, text-to-speech and speech-to-text Android client, backtracking “thoughts” for simple deduction, and user-defined skills as easy as a wiki article called Susy dreams. The system runs (downloadable server or hosted at dream.susi.com), already leverages 1.8 billion tweets for answers, supports session-scoped skills and pull-request review, and is actively used.
AI-written summary · about
chapters (13)
- 0:00Intro
- 2:43Strong AI vs Weak AI
- 5:02Cognitive Assistant
- 6:19Create your own skills
- 6:39Zeusie
- 10:53Conversation Routes
- 12:49UserDefined Skills
- 13:04Fixed Answers
- 18:50Cognitive Assistance
- 19:40Susy Dream
- 20:21Warnings
- 20:38Elon Musk
- 21:26Susie
related talks
44mSmartly AI: Cross Platform Voice Application Creation
29mBuilding an A.I. Assistant that Speaks Human
10mApp-Based Task Shortcuts for Virtual Assistants
51mBuilding an AI assistant that makes phone calls
31mAlmond: Crowdsourcing an Open, Programmable Virtual Assistant
15mDicio voice assistant — Libre, multilanguage and offline assistant for Android
23mPersonal OS: Build Your Army of 100 AI Interns
52mDesigning digital assistants for those who need it most