Dan Wahlin: AI coding agents make security, testing and human judgment matter more
DanWahlin · x · 2026-11-12
In the Code To Cloud podcast, Microsoft developer relations expert Dan Wahlin explains why engineering discipline matters more when agents write, deploy and open pull requests.
- Security first: never assume an agent's PR is fine; secrets in .env files are the classic failure
- Let agents build Azure/Kubernetes deployments, but demand edge-case analysis
- "Building 10 Golden Gate bridges at once": after everything ships faster, verification becomes the bottleneck
- Practices: stacked PRs, steering agents toward option B, e2e tests in the pipeline (not just on a laptop), deterministic automation to stop burning tokens
Takeaway: faster AI-driven development raises the value of testing, evidence and human judgment.
More from coding & agent
- Alchemy author demos Z3-backed Prove/Explore for TypeScript domain invariants — samgoodwin89 · 2026-11-12
- SteerCamp user racks up 1,182 agent memory items in 2 weeks with natural recall — pramodk73 · 2026-11-12
- 3 GitHub repos worth reading for the code: nanoGPT, tinygrad, llama.cpp — cneuralnetwork · 2026-11-12
- Reddit comparison of 7 agent messaging approaches: A2A, email, Slack, custom protocols — StrongCoffee85 · 2026-11-12
- Dev feeds new Claude a Call of Duty video, ships a playable browser tactical shooter — TAbrodi · 2026-11-12
- Second V8 patch lands: TypedArray indexOf/includes up to ~30x faster on Uint8Array — cnakazawa · 2026-11-12