AI agents - VoIP.ms Wiki

Check out our YouTube channel to watch our simple tutorials that will help you set up most of our features.

AI agents

From VoIP.ms Wiki

Revision as of 15:02, 9 June 2026 by Johann Jimenez (Talk | contribs)
Jump to: navigation, search
AI Agents (Beta)

Configure a Voice Agent to better handle your inbound calls.


This guide explains how to configure VoIP.ms AI Voice Agent to handle inbound calls within your existing telephony setup.

The AI Voice Agent acts as a virtual first point of contact, capable of answering calls, understanding caller intent, and responding based on the behavior you define during configuration. Proper setup is critical, as the agent's performance is directly influenced by how clearly its instructions, call flows, and routing logic are defined.

Important: The AI agent does not "learn" your business automatically. It operates based on the rules, prompts, and context you provide — making configuration a key factor in achieving accurate and reliable interactions.

Pricing

During the Beta Period, the pricing is $10 per agent per month. Afterwards, the standard price will be $25 per agent per month. Furthermore, per-minute pricing is $0.25/min. This includes STT, TTD, and LLM capabilities, so you get full functionality with no hidden costs.


Contents


Getting Started

To access the AI Agents feature, navigate to the VoIP.ms Customer Portaland click on AI Agents (BETA) in the navigation bar.

AI Agent - Dashboard.png

Creating an Agent

Click on Create New AI Agent.

IA agents 2.png

You will see two options:

  • Simple Voice Agent — A guided, step-by-step setup ideal for most use cases. Covers the essentials: voice, personality, purpose, call routing, and a knowledge base.
  • Advanced Voice Agent — Provides full control over the LLM model, TTS model, MCP servers, webhook tools, and custom parameters for complex integrations.
IA agents 3.png

Choose the option that best fits your needs. The sections below cover both paths in detail.

Simple Voice Agent

The Simple Voice Agent walks you through setup in a series of focused steps. Complete each step and click Next to advance.

Step 1: Basic Configuration

In this section, you will define the foundational details of your agent:

  • Agent Name — Assign a name to easily identify the agent
  • Description — Provide a brief explanation of the agent's purpose (eg. Support Agent Level 1)
  • Voice Selection — Choose from an extensive list of available voices
AI Agent - Basic Configuration.png
Note

Not all voices currently have preview functionality. This will be expanded in future updates.

Once completed, click Next to proceed.

Step 2: Personality and Tone

This section defines how your agent communicates. Based on your answers, VoIP.ms starts to build a prompt on your behalf in the back-end.

  • Language — Select the primary language for the agent (English, Spanish or French)
  • Tone and Interaction Style — Select how the agent should interact with callers. By default, simply go for Short and Concise.
  • Agent Personality — Use this field to provide additional instructions and context, such as: agent name, company name, behavioral guidelines, or communication style preferences. This input directly influences how the AI behaves during calls.
AI Agent - Personality and Tone.png

Step 3: Purpose & Communication

Here you need to define what the agent is designed to do and how it interacts at the start of a call:

  • Purpose of the Agent — Select the purpose of the agent anywhere from Troubleshooting Common Issues (Default), to Collect Basic Information.
  • Greeting Message — Define the initial message the agent will deliver when answering a call. This text will be spoken exactly as written, so ensure clarity and professionalism. For instance: Hi, this is Violet with Company X Support. I can help for most issues or connect you with a live engineer. What seems to be happening today?
AI Agent - Purpose and Communication.png
>>> To be changed.

Step 4: Call Routing & Transfer Options

This section allows you to configure whether the AI agent can transfer calls. If you do not want call transfers, simply click Next. Otherwise, to enable transfers, click the Allow AI agent to transfer calls to specific destinations checkbox, then add a label to describe your transfer scenario(s) (e.g., "Sales Team", "Support Line") or the name of the given person (e.g. "John Doe").

Then, add as many destinations as you wish from:

IA agents 11.png

Step 5: Knowledgebase

In this section, you define the information sources the agent will use to respond to callers. You can add links such as your website, a knowledge base, a FAQ page, or even a Wiki. Ensure all links are publicly accessible and readable.

Optional: Enable Allow web search for additional information when appropriate to supplement responses with live web results.

Once all sections are completed, click Create Agent to finalize the setup of your Simple Voice Agent.

IA agents 12.png

Advanced Voice Agent

The Advanced Voice Agent provides full control over your agent's underlying models, integrations (coming soon!), and behavior. Use this option when you need to select a specific LLM or TTS model, connect external tools via MCP servers or webhooks, or define custom parameters for complex use cases.

General Settings

In this section, you define the core configuration of your advanced agent:

  • Agent Name — Assign a name to identify the agent
  • TTS Model — Select the text-to-speech model for voice generation
  • Voice — Choose the voice used during calls. Please note that some voices currently do not have preview functionality. This will be expanded in future updates.
  • LLM Model — Select the language model that powers the agent's reasoning and responses
  • Background Audio — Optionally add background sound during calls
  • Language — Define the primary language of the agent (English, Spanish or French)
  • Description — Provide a brief summary of the agent's purpose
IA advanced 1.png

Call Routing and Transfer

This section allows you to configure whether the AI agent can transfer calls. To enable transfers, click the Allow AI agent to transfer calls to specific destinations checkbox, then add a label to describe your transfer scenario(s) (e.g., "Sales Team", "Support Line") or the name of the given person (e.g. "John Doe").

Then, add as many destinations as you wish from:

IA advanced 2.png

Knowledge Base

In this section, you define the information sources the agent will use to respond to callers. You can add links such as your website, a knowledge base, a FAQ page, or even a Wiki. Ensure all links are publicly accessible and readable.

Optional: Enable Allow web search for additional information when appropriate to supplement responses with live web results.

AI Agent - Advanced - Knowledge Base.png

Advanced Tools and Integrations

Extensions allow AI Agents to connect with external systems such as CRMs, helpdesks, databases, and internal business applications.

These integrations enable your AI agents to send information, retrieve data, and perform actions. Helping automate workflows and provide more personalized customer interactions.

Extensions are configured from the Extensions section within the AI Agent configuration page.

There are two types of integrations available:

  • Webhooks
  • MCP Servers

Each serves a different purpose and can be used independently or together depending on your workflow requirements.

Webhooks

Webhooks allow your AI Agent to send information to an external system when a specific event occurs during the call lifecycle.

In simple terms:

"When something happens, send this information to another system."

Webhooks are commonly used to:

  • Create support tickets
  • Log call information
  • Send call summaries
  • Update CRM records
  • Trigger internal workflows
  • Notify external applications

Webhook Configuration

The following fields are available when creating or editing a webhook.

Field Description Purpose
Name Internal identifier Unique system identifier used by the AI Agent to reference the webhook.
Friendly Name Display name Human-readable name displayed within the Portal.
Description Usage description Explains the purpose of the webhook and when it should be used. The AI Agent may use this information when determining whether to execute the webhook.
Trigger Condition Execution timing Defines when the webhook is triggered.
HTTP Method Request method Defines how data is sent to the destination system. Common methods include POST, GET, PUT, PATCH, and DELETE.
URL Destination endpoint Web address that receives the webhook request.
Parameters Data fields Information sent to the external system.


Parameters

Parameters define the information that will be sent to the destination system.

For each parameter, you can specify:

  • A dynamic value generated by the AI Agent
  • A fixed value that is always sent
  • A default value used when information is unavailable

Dynamic Values

If the Value field is left empty, the AI Agent will attempt to populate it using information collected during the conversation.

Examples include:

  • Customer name
  • Email address
  • Phone number
  • Call reason
  • Issue description

Fixed Values

A fixed value will always be sent with the request.

Example:

  • priority = normal

Default Values

A default value will only be used if the AI Agent is unable to determine the information from the conversation.

Example Parameter Configuration

Parameter Value Result
subject AI generates a subject based on the conversation.
description AI generates a call summary.
customer_email AI attempts to capture the caller's email address.
priority normal Always sends "normal" as the priority value.


Webhooks.png


MCP Servers

MCP (Model Context Protocol) Servers allow AI Agents to interact with external systems in real time during a conversation.

Unlike webhooks, which send information in a single direction, MCP Servers establish a live connection that enables the AI Agent to query external tools and retrieve information while speaking with the caller.

In simple terms:

"Connect the AI Agent to an external system so it can retrieve information and perform actions during the call."

Common MCP Use Cases

MCP Servers can be used to:

  • Retrieve customer information from a CRM
  • Check order status
  • Verify product availability
  • Access account information
  • Search internal databases
  • Execute business-specific workflows

Because these actions occur during the conversation, the caller can receive immediate responses without requiring manual intervention.


MCP Server Configuration

The following fields are available when configuring an MCP Server.

Field Description Purpose
Name Connection name Friendly name used to identify the server connection within the Portal.
Transport Type Connection method Defines how the AI Agent communicates with the MCP Server. Supported options include SSE and HTTP.
URL Server endpoint Address of the MCP Server provided by the connected platform.
Auth Headers Authentication credentials Security information required to access the external system.

Authentication Headers

Authentication Headers are commonly used to grant access to external systems.

Examples include:

Header Example Value
Authorization Bearer abc123...
X-API-Key your-api-key

These values are typically provided by the technical team responsible for the connected application.

Additional Notes

Some MCP implementations support advanced initialization options that may be configured through the API.

These settings are typically used for custom integrations and advanced deployment scenarios.

MCPs.png


Communication Settings

In this final section, you can define how the agent interacts at the start of a call and can also build the entire prompt :

  • Greeting Message — Set the initial message delivered when answering a call
  • System Prompt — Define the core instructions that guide the agent's behavior and responses (see AI Prompts examples here)
IA advanced 5.1.png

Linking Your Agent to a DID

Once your agent is created, you need to assign it to a phone number (DID) so it can start answering inbound calls.

  1. Go to DID NumbersManage DIDs in the customer portal
  2. Click Edit on the DID you want to route to the agent
  3. In the routing destination section, select AI Voice Agent from the dropdown
  4. Choose the specific agent you created from the list
  5. Click Apply

Your AI Agent will now answer all inbound calls to that DID!

AI Agent - Manage DIDs.png


Frequently Asked Questions

Q: Can the AI agent handle outbound calls?

A: No — AI Agents are designed for inbound calls only. They answer calls made to your DID; they cannot initiate outbound calls.

Q: What languages does the agent support?

A: English, Spanish, and French.

Q: What is the difference between Simple and Advanced?

A: The Simple Voice Agent guides you through a fixed set of steps and covers everything most users need: voice, personality, purpose, call routing, and a knowledge base. The Advanced Voice Agent exposes the underlying LLM and TTS model selection, background audio, and (coming soon) MCP servers and webhook tool integrations for more complex use cases.

Q: Can I have multiple agents?

A: Yes — you can create as many agents as needed and assign each one to a different DID.

Q: Can I edit my agent after creating it?

A: Yes — from the AI Agents dashboard you can edit, duplicate, or delete any agent at any time.

Q: What happens if the agent cannot answer a question?

A: The agent will let the caller know it does not have that information. If call transfers are configured, it can route the caller to a human agent instead.

Q: Does the agent answer calls 24/7?

A: Yes — once assigned to a DID, the agent answers all inbound calls at any hour.

Q: Does the AI agent work with my existing DID?

A: Yes — any eligible DID can be routed to an AI Agent from DID Numbers → Manage DIDs.

Q: What does "(Beta)" mean?

A: The feature is in active development. Some capabilities — such as Advanced Tools and Integrations — are not yet available, and behaviour may change before the general release.

We will keep populating this FAQ as questions come along. Thanks in advance for your feedback, support and questions!

Need Help or Got Feedback?

Support resources

When reporting an AI Agent issue, include:

  • The name of the agent and the DID it is assigned to
  • The date and time of the call, including the UniqueID found in the CDR
  • A description of the behaviour observed vs. what was expected
Personal tools
Namespaces
Variants
Actions
VoIP.ms Wiki
Guides 🇨🇦
Guías 🇲🇽