Personalized EB-1A Guidance
summary
A mobile-first conversational system for e1a guidance that grounds answers with retrieval augmented generation and a document corpus, backend in nodejs with JavaScript, iOS client built in Swift SwiftUI following TCA. Exponential backoff cuts API failures and reduces response time from ~10–15s to ~5–7s. Produces summarized answers plus bulleted action items; currently iOS-only, uses an unnamed LLM, and plans include Android/web, voiceover, globalization and autogenerated personalized to‑dos.
AI-written summary · about
chapters (9)
- 0:00intro
- 0:14preamble
- 2:10overview of the intelligent conversational system
- 4:02background and core tenets of the ev1a bot
- 5:55artificial intelligence and retrieval augmented generation
- 8:26mobile application development and client-server interaction
- 10:49user experience and design considerations
- 14:07challenges and future work
- 18:31conclusion and final thoughts
related talks
38mCreate the next generation of user experiences with text generation
28mBridge the AI knowledge gap with Modern Web Guidance
01h 04mBuilding Expertise Into Generated Conversations
17mBuilding Effective Voice Agents
33mYour e-commerce bot
20mAnswers. Agents. AGI.
01h 04mBuilding Expertise Into Generated Conversations
36mGENAI-Powered chatbot to increase team efficiency