Grady Booch: Coding agents write code well, but architecture still needs human judgment
Grady_Booch · x · 2026-10-02
Software engineering pioneer Grady Booch argues human-led architecture matters even more in the era of generative AI coding agents. Architecture, he notes, is the set of significant design decisions — significance measured by cost of change — and while agents are decent at turning prompts into code, they struggle to balance static and dynamic forces on a system. That trade-off judgment is where human experience still shines.
More from coding & agent
- VLM Run launches SystemOne API: typed, calibrated answers in one forward pass — spillai · 2026-10-02
- The tail breaks at scale: VLM p90 latency 209ms vs OCR's 1453ms — spillai · 2026-10-02
- Document classification bake-off: VLM is ~7x more token-efficient than OCR — spillai · 2026-10-02
- sindresorhus disables external PRs on all repos, citing AI: 'open source as we knew it is over' — DanielLockyer · 2026-10-02
- AWS revamps onboarding: no credit card, $200 free credits, hard spend caps, one-prompt agent setup — danbri · 2026-10-02
- Eve adds Link agent wallet support: give your AI agent payments in a few lines of code — jeff_weinstein · 2026-10-02