Meta’s RADAR: automate low-risk review with calibrated risk
Industrial study of RADAR at Meta—risk calibration and selective automation to reclaim reviewer time on low-risk changes.
Source: Automating Low-Risk Code Review at Meta: RADAR, Risk Calibration, and Review Efficiency
Not every diff deserves the same human scrutiny. Meta’s RADAR paper is about routing: calibrate risk, automate the low-risk slice, and spend scarce reviewer attention where it pays off.
The interesting engineering is the risk model and the feedback into review efficiency—not a claim that AI replaces review. It is factory thinking: classify the change, apply the matching gate.
Why it matters for code review
Software factories need differential review, not uniform LLM spam on every PR. RADAR is a concrete industrial pattern for attaching automation to low-risk paths while keeping humans on high-risk ones—exactly the control plane most agentic SDLC setups are missing.