Glossary
Response latency
The time between the visitor’s question and the start of the agent’s answer. Under a second the conversation flows; past three, abandonment climbs fast, because the wait becomes noticeable.
Latency is measured from the end of the visitor’s question to the first token of the answer: not to the end of it, because a streamed answer starts being read before it finishes.
The thresholds are behavioural, not technical: under a second reads as instant, around two as thinking, past three or four people switch tabs. In voice the tolerable window is roughly half that.
Most of the budget goes to retrieval and generation, which makes this largely a design decision: how many sources to search, how many checks to run, whether to stream. Chasing the last 200ms usually costs more than it returns; closing the gap between four seconds and one is what actually moves conversion.
Start the conversation today
Launch an AI agent that qualifies leads and books meetings around the clock. Live in under ten minutes.
A guided demo · your agent live in under 10 minutes