Don't Build Agents You Can't Explain
addyosmani · x · 2026-07-16
In his closing keynote, Addy Osmani emphasized: do not build Agents you cannot take responsibility for explaining.
Key points include:
- Roles/titles < system ownership: Rather than giving someone an "AI-flavored" job title, it is more important to clarify who is accountable for the system's outcomes.
- Quoting Mitchell Hashimoto: when the "speed of generation" exceeds the "speed of understanding," the scarce resource becomes evidence-based judgment.
- His proposed new rule is: "Explain it, or don't ship it."
The post focuses less on specific tools and more on accountability, judgment, and explainability in Agent products and team practices.
Related event: Addy Osmani: Engineers Must Be Accountable for Agent Systems(3 posts)→
More from coding & agent
- AIE Code Summit 2026 opens applications: 800 seats, keynotes streamed free — trq212 · 2026-10-27
- Decade-long cloud architect: enterprise AI architecture is far more than bolting on an LLM — kashifmanzoor · 2026-10-27
- Analysis of thousands of coding sessions: 40-89% of files agents read are never used — rseroter · 2026-10-27
- TRL adds Score Centering, a simple fix for RLHF training instability — lvwerra · 2026-10-27
- Precompute golden answers and match at runtime for exact wording control — colinmcnamara · 2026-10-27
- Dev argues there's still alpha in working with your coding agent interactively — alex_frantic · 2026-10-27