Risk Scoring and Silence Management

Per-turn risk assessment and caller inactivity handling during live voice calls.

During a live voice call, the platform continuously evaluates two things: how risky the current conversation is, and whether the caller has stopped responding. These two systems - the risk scorer and the silence monitor - operate independently but both feed into the agent's escalation logic.

Risk Scorer

The voice agent computes a composite risk score on every conversational turn. The score combines three signals, each weighted to reflect its contribution to overall call health:

Signal	Weight	What It Measures
Emotion	40%	Negative valence combined with high arousal, deteriorating emotional trend, barge-in frequency, and consecutive short responses from the caller
Loop detection	30%	How many times the agent has revisited the same state. Repeated state visits suggest the conversation is going in circles without making progress.
Duration	30%	Time elapsed relative to the expected call length. Risk ramps after the expected duration (default 5 minutes) is exceeded.

Risk Levels

The composite score maps to four levels:

Level	What Happens
Normal	Agent operates with full autonomy within its context graph.
Monitor	Internal flag raised. Agent behavior unchanged, but the call is flagged for closer post-call review.
Alert	Available operators receive a notification. Agent continues but with increased caution.
Escalate	Automatic escalation to an operator if one is available.

Per-State Threshold Overrides

Individual context graph states can override the default thresholds. A medication verification state might have a lower escalation threshold than a general scheduling state, because errors in that context carry higher clinical risk.

This means the system adapts its sensitivity to the stakes of the current conversation phase. Routine scheduling operates with standard thresholds. Clinical data collection tightens them.

Silence Monitor

If the caller stops speaking, the system needs to determine whether they are thinking, distracted, or have left. The silence monitor manages this with exponential backoff:

1. After 10 seconds of silence, the agent asks "Are you still there?"

2. If no response, the next check-in waits 20 seconds

3. Third check-in at 40 seconds

4. After three unanswered check-ins, the agent ends the call with a message offering to have someone call back

Staleness and Interruption

Check-in utterances have a 5-second staleness window. If the caller speaks during the check-in filler, the filler is discarded and the conversation continues normally. The silence monitor resets whenever the caller speaks.

This prevents awkward overlaps where the agent says "Are you still there?" just as the caller starts talking. The staleness window ensures the check-in is only delivered if the caller is genuinely silent.

The silence monitor and risk scorer are independent systems. A call can escalate due to high risk while the caller is actively speaking, or the silence monitor can end a call that has normal risk scores but no active participant.