WEM Real-Time Adherence - 'INVALID_AGENT_STATE' - Random Intermittents

INVALID_AGENT_STATE - this keeps popping up in the WEM Real-Time Adherence view. It’s not consistent, so it’s hard to track, but it’s causing inaccurate reporting on agent states - mainly showing agents as ‘NOT READY’ when they’re clearly in ‘AVAILABLE’.

The WEM engine seems to be getting confused about the agent’s ACTIVITY. We’re seeing this mostly on agents using the Zoom Contact Center integration. It’s not happening with everyone, and it doesn’t seem tied to a specific skill or team, but it’s enough to cause problems with SERVICE LEVEL reporting.

IF (AGENT_STATUS == AVAILABLE) THEN
UPDATE (WEM_STATUS == AVAILABLE)
ELSE
UPDATE (WEM_STATUS == NOT_READY)
ENDIF

It feels like a sync issue between Zoom and the WEM engine - the state change isn’t propagating correctly. I checked the ARCHITECT flow - specifically the agent state updates - and nothing looks obviously wrong. The flow is pretty straightforward, just updating the agent state based on Zoom events.

Here’s what we’ve checked already:

  • Genesys Cloud region: US West 2
  • WEM SDK version: 2.1.4
  • Zoom Contact Center integration version: latest (updated last week)
  • Checked AGENT_PROFILE for correct time zone settings
  • Reviewed the REAL_TIME_ADHERENCE API logs - nothing helpful, just the INVALID_AGENT_STATE errors.
  • Confirmed the QUEUE_ROUTING_CONFIG is pushing the correct agent state updates.

It’s just… random. Sometimes it works fine all day, other times it’s happening every few minutes. We’ve been pulling reports hourly, but the data is unreliable because of this.