Intelligence as Infrastructure: What Developers Must Learn to Build with AI
summary
Argues developers must move from simple chatbots to proactive UIs, process automation and agentic systems; shows a POC that filters roughly 70% of file-review work using pipelines and human audit fallback, demos an extended-reasoning fan-out + majority-voting pattern (built with GitHub Copilot and ChatGPT) to amplify model thinking, recommends routing agents instead of vector DBs for fast RAG prototypes, and describes an experimental agentic runtime (YAML/ZAML emulator) that partly works but is costly and still tends to chatter rather than reliably act.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 1:01Chatbots to proactive AI
- 5:33RAG without vector DBs
- 7:16Process automation with agents
- 10:47AI reasoning with evolution
- 13:36Prompts as code
- 16:43Refactoring AI steps
- 18:27AI agents for onboarding
- 21:13Agentic runtimes
- 27:53Limits of agentic AI
- 29:59AI as shared compression
- 37:43Philosophy of AI scaling
- 38:25Q&A
related talks
59mFrom Pair to Peer Programmer
51mCode Smarter, Not Harder: AI-Powered Dev Hacks for All
47mAI and the Future of Software Development
13mMeet the Speakers
01h 16mAI in Software Development: Boosting Efficiency
14mThe 4 Patterns of AI Native Development
46mPanel Discussion: AI's Impact on Developers
27mWill AI (Finally) Replace Developers?