A 311 Foundry deployment just launched and product owners need live quality signals after go-live. Which capability should they enable?

Enabling continuous monitoring in Foundry to watch live quality signals after launch is correct because pre-launch offline evaluations cannot capture production drift, traffic mix, or content changes that appear only after residents use the system. A 311 Foundry deployment that just went live needs ongoing visibility into quality metrics—groundedness, relevance, safety proxies, latency, and error rates—so product owners can detect regressions quickly. Continuous monitoring turns GenAIOps from a one-time gate into sustained civic service quality management. Batch evaluation in Foundry to rate live traffic for quality signals after launch scores sampled live traffic in periodic offline batches, so it cannot provide the continuous live visibility product owners need right after go-live. Latency dashboards in Foundry to track live traffic health signals after launch monitor speed and operational health, not the live quality signals the stem asks for, so they leave product owners blind to content and relevance regressions during normal resident traffic. Auto retraining in Foundry to update models from live quality signals after launch updates models using those signals and does not give product owners live quality signals after launch, so it cannot replace continuous monitoring for catching short-lived incidents. Exam caveat: continuous monitoring is the production counterpart to pre-deploy evaluation. Batch evaluation in Foundry to rate live traffic for quality signals after launch scores samples in retrospective batches; Foundry continuous monitoring watches live continuously after launch. The stem asks for live quality signals post go-live—choose continuous monitoring. Operational check: enable Foundry continuous monitoring for the 311 app, define alert thresholds on quality and safety signals, route alerts to product owners and on-call, review dashboards in a recurring ops cadence, and open incident or rollback workflows when live metrics breach municipal SLOs.