Google study: devs expect agents to perform as-is while juniors grow via review
rseroter · x · 2026-11-07
RDEL #168 covers Google research on what behaviors developers actually want from AI coding agents as teammates.
Key points:
- Researchers analyzed 91 agent rule files and interviewed 15 experienced Google developers
- Central contrast: participants expected junior engineers to grow through code review, but expected agents to "perform as is," improving only through the user's prompts and context
- Benchmarks like SWE-bench measure whether patches pass tests, not whether agents follow conventions, ask before risky changes, or improve from feedback
- Rule files grow like onboarding notes, one correction at a time after each agent mistake
Author rseroter notes the data is about a year old but likely still applies.
More from coding & agent
- PSA to coding agents: keep CSS :has() rules tight to avoid huge style recalcs — jh3yy · 2026-11-07
- iOS 27.3 reveals Siri's memory graph: memory atoms fed by Fastpass and agent recipes — alexisgallagher · 2026-11-07
- Indie dev adds AI agent contact editing with per-change undo and approval gates — SeanOliver · 2026-11-07
- Designing an MCP server for 60+ financial datasets: entity keys, per-connector auth, context budgets — vianexus_official · 2026-11-07
- zeeg to livestream Junior in action, breaking down technical details of the agent — zeeg · 2026-11-07
- Sketching musical interfaces with ASCII art using Gemini in Google Antigravity — alexanderchen · 2026-11-07