Designing AI-Intensive Applications
summary
Argues AI apps shift from 1:1 to 1:100 to 0:n human input:LLM-call ratios, names LM OS and LM SDLC as candidate standard models, and highlights MCP and agents SDK changes. Describes a production SPADE pipeline for AI News — scrape, plan, analyze, deliver, evaluate — that runs daily with multiple scrapers, thousands of LLM calls and 70,000 daily readers, and says commodity tooling covers early SDLC while real engineering work is evals, security and orchestration.
AI-written summary · about
chapters (12)
- 0:00Conference Welcome and Overview
- 0:42Conference Logistics and Growth
- 1:47Audience Preferences and Survey
- 2:22Innovations in AI Engineering (MCP and Chatbots)
- 2:58Evolution of AI Engineering (Past Talks)
- 3:50Simplicity in AI Engineering
- 4:17AI Engineering as a Developing Field
- 5:23Seeking the "Standard Model" in AI Engineering
- 6:02Candidate Standard Models in AI Engineering
- 9:26Human Input vs. AI Output (AI News Example)
- 11:05SPADE Model for AI-Intensive Applications
- 12:29Call to Action for Conference Attendees
related talks
36mThe Best AI Strategy? Don’t Build AI 🤯 Build AI-Ready Software
27mWill AI (Finally) Replace Developers?
30mUsing AI to Create Better Requirements and Designs
19mThe New Application Layer
33mEngineering AI Applications: A Conversation with AI Leaders
26mAI & SRE: Embracing the Ultimate Frontier of Automation
21mUsing the Power of APIs and Domain Driven Design to Embrace AI [...]
26mAI-augmented SDLC: Rethinking how we deliver software