Bring the Power of AI to Your Application
summary
Explains why large language models bullshitting is the core risk and how user flows mitigate it: choose a smaller model for low-latency inline suggestions, prefer all-at-once or HTTP2 streaming depending on UX, and use a multi-step Copilot Workspace flow (specification → plan → code) so humans approve changes. Shows retrieval can be simple SQL retrieval not only vector DBs, cites real failures (fabricated legal precedents, replaced clinicians) to justify defensive, editable outputs.
AI-written summary · about
chapters (7)
- 0:00AI and user experience
- 1:52AI application development
- 9:14Designing AI apps
- 17:17Structured AI workflows
- 19:24Beyond chatbots
- 20:05AI adoption challenges
- 22:59Resources and closing
related talks
46mRedefining User Experience with Large Language Models
48mMastering the AI Shift: Adaptive Strategies for Software Evolution
01h 02mArchitecture Patterns for AI-Powered Applications
26mMastering Large Language Models: Prompt Engineering for Precision & Performance
56mHow To Supercharge Your Career And Business In The Era Of AI
32mBeyond Bolt-On AI: A New Way to Build Software
06m7 practical tips for safer AI and language models
25mHuman-Centered AI: What I've Learned From Building Two LLM-Powered Products Thi...