Beyond the code: Rediscovering the timeless skills of software engineering in the AI era
summary
Tries to get an AI to build an e‑commerce site and finds it produces usable fragments but not a full product because of context degradation (feeding 300 pages of specs makes performance drop). Demonstrates using cloud code’s plan mode and agentic workflows to break work into focused tasks, warns that a 100‑step automated decomposition accrues failure risk (probabilistic accumulation — e.g., 99% per step → ~36% overall). Shows a concrete schema mistake (price embedded in product) that breaks promotions, so human checkpoints and architectural trade‑offs remain essential.
AI-written summary · about
chapters (11)
- 0:00AI in software development
- 1:47Breaking down problems
- 4:04Context management in LLMs
- 7:07Techniques for AI coding
- 10:00The role of agentic workflows
- 11:54Focusing on context-free tasks
- 13:20Probabilistic accumulation
- 14:05Internal system architecture
- 16:16Fundamentals of design
- 18:25Choosing architectural patterns
- 21:24The future of software engineering
related talks
28mBeyond the code: Rediscovering the timeless skills of software engineering in the AI era
45mThe Rebirth of Software Craftsmanship in the AI Era
27mEngineering in the Age of AI: Making AI Work for Us, Not Replace Us
42mA Developer's Search for Meaning: Thriving as AI Transforms Our World
41mBeyond Code: The Essential AI Skillsets for Development Teams
54mAI is Coming - Now What? Architecting the Future of Software
26mbuilding reliable AI systems in production
29mPairing with the machine: How AI Agents are rewriting the rules of software development