Sysevo

How many agents do you actually need?

Standard Erlang C workforce math, the same formula call centers have used for decades. Put in your call volume and target service level, and see the staffing number, plus what changes when an AI voice agent removes the queue entirely.

Talk time plus any after-call work, per call.
e.g. answer 80% of calls within this many seconds.
Agents needed7to hit 80% of calls answered within 20s
Occupancy57%agent utilisation
Est. wait if delayed11saverage, when queued
What an AI voice agent changes

An AI voice agent has no queue. Every caller is answered in under 2 seconds, whether that’s 1 call or 100 at once — this entire staffing calculation stops applying.

How this is calculated

This uses the Erlang C formula, the standard model call centers use for staffing math. Offered load (in Erlangs) is calls per hour × average handle time, converted to a rate. The calculator then finds the smallest number of agents where the probability of a caller waiting longer than your target answer time keeps you at or above your target service level.

Erlang C assumes calls arrive randomly (a Poisson process) and callers don’t abandon the queue. Real call patterns are burstier than that around peak hours, so treat the output as a solid planning baseline, not an exact guarantee — most teams staff slightly above the raw Erlang C number for that reason.

Roughly 10,392 calls/month at this call rate, assuming an 8-hour day, 5-day week.

Related reading