How it works
What happens between the ring and the answer.
There is no magic step in here. A call arrives, an agent that knows your business answers it, and what was said is written down. The parts worth understanding are where its knowledge comes from and what it does at the edge of that knowledge, because that is where an AI answering your phone either earns its place or embarrasses you.
The sequence
Five stages, in order.
Stage 1
The call arrives.
Your number is a Twilio number that points at agentmomo. The agent knows which business it is answering for from the number that was dialled, so nothing has to be passed to it and there is nothing for a caller to select. On the web it is the same agent behind a chat box on your own pages, with no login and no cookie wall in front of it.
Stage 2
It answers as your business, from your material.
Before it goes live we load what you already have: your hours, your services, your policies, the answers you give twenty times a week. The agent retrieves from that, and only from that. It is not answering from the open internet, so it does not develop opinions about your business that you do not hold.
Stage 3
It stays inside rules you set.
A profile decides what the agent is allowed to say. The practice profile, which is the one answering the demo on this site, is forbidden from giving clinical advice, from quoting a price, from naming insurance carriers and from claiming you take online bookings. An agent that refuses to invent a price is the product working. The refusal is the feature you are buying.
Stage 4
When it cannot answer, it gets a person involved.
It says it does not know, takes the caller’s name and number and what they wanted, and hands that to you rather than guessing. Every profile can capture a lead. That is the honest bottom of the funnel: the caller is not left with a wrong answer and you are not left without the call.
Stage 5
You read what was said.
Every conversation, on every channel, is stored as text and readable afterwards in the console. There are no call recordings anywhere in the system, only transcripts. When a customer tells you your system said something wrong, you can go and look at what it actually said, which is the reason this is written down at all.
The three channels
One agent, three ways in.
The same brain and the same knowledge answers on all three. They differ in what the medium allows, not in what the agent knows.
Phone
Inbound calls over Twilio, bridged to a realtime voice model. Every session has a length cap with a spoken wind-down before it ends, so a stuck call cannot run away with your phone bill.
SMS
A real channel with its own console surface for reading and replying, plus a control to take the conversation over from the agent. A new number has to clear carrier A2P registration first, which is a carrier timeline measured in weeks and not something we can shorten.
Web chat
Embedded on your own pages, unauthenticated, streaming. The visitor on your website gets the answer the caller would have gotten.
The honest edge
It does not know that the caller is the person who chatted yesterday.
Continuity holds within a conversation and within a channel. It does not carry across channels: a caller and a web visitor are two records, even when they are one person. We say this on every page that touches the subject rather than letting the phrase "one memory" quietly imply something we have not built. Everything is stored in one place and readable by you; the agent is the one who does not join them up.
Next
It is quicker to use it than to read about it.
The agent on the home page is running every stage above except the one where a person at your business reads the transcript afterwards.
Try the demo