Voice Agents
Setting up your first Voice Agent
The decisions that shape a voice agent, from the job it does to the systems it touches and the moment it hands over to your team.
A voice agent is defined by four things: the job it does, what it knows, what it can do in your systems, and when it hands over to a person. Your Sarj deployment team configures each of these with you. This article explains the choices so you can come to those sessions ready.
Define the job
Start with one conversation your team handles at volume and understands well. A focused first agent goes live sooner and gives you clean results to build on.
Inbound support
Answer the questions behind most calls, such as order status, balances or appointments.
Outbound follow-up
Call leads back, send payment reminders or confirm bookings.
Give it knowledge
Policies and answers
The agent answers from the material you approve: policies, product details, fees and service hours. It does not improvise beyond it.
Customer context
Connected to your CRM, the agent knows who is calling and what they have open. See Connect your CRM.
Let it act
Decide which actions the agent may take on its own, such as rescheduling a delivery, booking an appointment or logging a promise to pay, and which always need a person. See CRM & ERP Actions.
Decide when it hands over
Some requests should always reach a person: complaints, vulnerable customers, anything outside policy. Set those rules before launch. See Agent Handoff Rules.
Connect the phone line
Sarj answers your existing numbers through its telephony SDK or standard SIP, so you choose which calls reach it while your current platform keeps handling the rest. See Telephony SDK & SIP.