AI Voice Agent
It answers, and the record is written before the call is over. Calls reach it from a queue the way they reach a person, on inbound DIDs and outbound campaigns from one configuration, and what it captures lands mid-call, not in a summary afterward.
Voice agent: Front desk
may writeGood morning, you have reached Meridian. Who were you after?
Product figures from the platform’s own defaults - not customer averages
How it works.
It waits on the queue like everyone else
An inbound DID rings it the way a queue rings a person, and an outbound campaign hands it a connect the same way. One configuration covers both directions, and the call stays where it was placed.
Hear, decide, answer
Voice activity detection marks the end of a turn, streaming speech to text puts it into words, a model with tool use reads the conversation so far and chooses what to do, and text to speech says it. The audio runs on the same media path as every other call.
It can do things, not just say them
It can transfer to a queue, set a disposition code, capture a field onto the lead, or end the call. Each of those is a write made while the call is up, not a summary assembled afterward, which is the line between this and a menu with better prose.
The trail you already read
Captured fields land on the record during the conversation, the disposition is one of the codes your people pick from, and every session keeps its full transcript for review call by call.
Two moments in every call.
Every call passes through the same seven. AI Voice Agent is the lit ones, and everything either side of it is a different page in this category.
- 01Number
the line it arrives on or leaves from
- 02Route
where the call is sent and to whom
- 03Dial
placing it, and how many at once
- 04Connect
the moment a person picks up
- 05Talk
while the conversation is happening
- 06Wrap
the code, the notes, the next step
- 07Record
what is kept, and where it is kept
The specifics.
8 facts- Tools
- transfer(queue), set_disposition(code), capture_field(key, value), end_call
- Pipeline
- Voice detection, streaming speech to text, a model with tool use, then speech
- Direction
- One configuration answers inbound DIDs and qualifies outbound campaigns
- Handover
- Warm transfer to a person on a queue, carrying the transcript and the fields
- Fields
- Written onto the lead mid-call, under keys you name when you build it
- Setup
- Greeting, objective, the questions it asks, the fields it captures, and the voice
- Not quoted
- No latency figure, language count, carrier name or accuracy rate is published
- Not the same as
- A menu that reads options and waits for a keypress is the auto-attendant. This one holds a conversation and writes while it does
What runs alongside it.
Volume calling is several controls working at once, not one dial turned up. These are the others.
More in Outbound & Compliance
2 capabilitiesBroadcasting, voice AI, QA scoring, calling rules and number health.
Nobody dials and no seat is paid for while the list gets through. One message reaches all of it, a machine takes the drop, a keypress is the route to a human.
Nobody answers a number that says Spam Likely, and the label lands before anyone tells you. Spam and CNAM status per DID, a warm-up ramp, a rest threshold.
The rest of the platform.
Five more categories, all on the same record and the same bill. Each card names three of its capabilities, so you can tell from here whether it is worth opening.
Outbound & Compliance
See ai voice agent on your own floor.
Thirty minutes, your numbers and your data. We will set ai voice agent up live and you can decide from the thing itself rather than from this page.
14-day trial · no card · migration included