Salesforce Decode
Salesforcedecode
Back to questions
AgentforceAdvancedresiliencedrfallbackavailability

Design disaster recovery and fallback when Agentforce is unavailable

Real World Scenario

During a Salesforce incident, Agentforce stopped responding but voice and chat channels remained open. Customers received infinite typing indicators for 45 minutes.

Expected Answer

• Implement health checks on agent endpoints with automatic fallback to static FAQ or queue-to-human • Configure channel-level contingency: IVR message, email auto-reply, or simplified bot with no LLM • Document RTO/RPO for agent services separately from core CRM—know Einstein dependency chain • Run game days simulating Agentforce degradation and measure operator response • Maintain offline playbooks for contact center supervisors to disable bot routing quickly • Communicate status page integration so customers know alternate contact options • Post-incident: add synthetic probes alerting before customer volume spikes

Follow-Up Questions & Answers

Click to expand — each follow-up includes a direct, interview-ready answer

Direct answer: Implement health checks on agent endpoints with automatic fallback to static FAQ or queue-to-human Also consider: Configure channel-level contingency: IVR message, email auto-reply, or simplified bot with no LLM In practice: Document RTO/RPO for agent services separately from core CRM—know Einstein dependency chain Optimize for scale and operational observability.

Architect Perspective

AI availability is not CRM availability. Design graceful degradation as first-class—customers forgive outages with clear messaging; they do not forgive silent bots. Keep a one-click kill switch for agent routing in runbooks.