You must roll out a major change to a multi-agent orchestration graph (new planner prompt, updated tool schemas, and a new subagent). Leadership wants limited user blast radius, fast rollback, and production traffic validation—not only lab evaluations. Which release methodology best fits?
Select an answer to reveal the explanation.
Short Explanation
Correct answer: C. Canary releases expose a small slice of real traffic to the new multi-agent configuration, enabling early detection of orchestration regressions while preserving quick rollback. Big-bang weekend cuts (A) maximize blast radius. Laptop-to-prod (B) skips environment gates. Permanent unaudited forks (D) create configuration drift. Pair canaries with evaluation metrics, trace correlation, and automated rollback triggers.
Full Explanation
AI-500 covers release methodologies including DTAP, blue/green, and canary for multi-agent deployments. Agent changes are not purely binary code swaps—prompt, tool, and subagent interactions can fail only under live distributions. Canary routing (by percentage, cohort, or region) limits impact while capturing production traces and quality scores. Define success criteria: tool error rate, human override rate, groundedness, latency, and cost. Option A is high risk for agentic systems. Option B violates multi-environment strategy. Option D without governance becomes dual-running chaos. Blue/green remains valid when you need instant environment swap; canary is preferred when you specifically need progressive production validation. Codify promotion and rollback in CI/CD rather than manual portal clicks.