Skip to main content
The AI Agents view is the main screen of the Connect App. It shows all agents assigned to your account and provides tools for creating new agents, editing existing ones, and managing agent settings.

​
Agent List

Alongside the agent list, the toolbar offers organization-level pages: Guardrails, Custom Voices, Knowledge Bases, Email Templates, and — once an outbound agent exists in the organization — the Outbound Suppression List, a do-not-call list checked before every outbound call (see Outbound Compliance). The agent list displays all AI phone agents with two columns:
  • Account — The agent’s name
  • Extension — The phone extension assigned to the agent (if applicable)
The list is sortable by clicking the Account column header. A downward arrow (▼) indicates the current sort direction. [Screenshot: AI Agents list view showing Account and Extension columns]

​
Searching Agents

Use the search bar at the top of the agent list to filter agents by name, extension, site, or description. Type your search term and the list will filter in real time.

​
Agent Row Actions

Each agent row displays five action icons on the right side when you hover over it:
  • Integrations (gear icon) — View and configure available integrations for this agent
  • Usage Analytics (magnifying glass icon) — View usage metrics and call statistics for this agent
  • Test (phone icon) — Open a WebRTC test client to have a live voice conversation with the agent; for an outbound agent, also place a test call to a phone number
  • Edit (pencil icon) — Open the agent editor to modify settings and configuration
  • Delete (red X icon) — Remove the agent permanently
[Screenshot: Agent row with action icons visible on hover]

​
Creating a New Agent

Click the New button in the top-right corner of the agent list to launch the new agent wizard. The wizard guides you through setup in several steps.
1

Choose Extension

Enter an unused extension number to assign to the new agent. This is the internal extension that will be used to reach the agent within the phone system.[Screenshot: Choose Extension step]
2

Select Caller ID

Choose a Caller ID (DID) from the dropdown. This is the phone number that will appear as the Caller ID when the agent makes outbound calls. The dropdown is populated with available numbers from your account.[Screenshot: Select Caller ID step]
3

Name and Notification Email

Enter an Agent Name — a descriptive name that identifies the agent’s purpose (e.g., “Front Desk”, “After Hours Support”). Then enter a Notification Email where agent notifications will be sent.[Screenshot: Create New Agent step with name and email fields]
4

Choose a Template

Select from one of the ready-to-use agent templates to get started quickly. Each template comes pre-configured with instructions and functions suited to a specific use case:
  • Receptionist Agent — Answers questions, screens calls, and transfers to appropriate destinations
  • Appointment Scheduler — Schedules appointments and takes messages according to instructions
  • Support Agent (CRM) — Connects with Zendesk, Freshdesk, or MCP Server to handle customer support inquiries
  • Blank Slate Agent — Start from scratch with a basic agent template
  • Message Taker Agent — Answers questions and takes messages that are emailed with the full call transcript and recording
  • Translator Agent — Interprets a call between two people who do not share a language, English with Spanish or French (see Translator Agent)
[Screenshot: Select a Template screen]
5

Template Setup Wizard

After selecting a template, a guided wizard walks you through the key settings for that template type. The steps vary by template but typically include:
  • Business name — Enter the name of your business so the agent can refer to it during calls
  • Greeting — Customize what the agent says when it answers calls (a default greeting is pre-filled based on the template and your business name). This step can be skipped.
  • Template-specific options — For example, the Receptionist Agent template asks whether to enable call transfers and shows available transfer destinations
[Screenshot: Template setup wizard - greeting step]
6

Transfer Destinations (Receptionist Agent)

If you selected the Receptionist Agent template and enabled transfers, this step shows the available transfer destinations. These are automatically populated from the NetSapiens directory — any user with “announce in directory” set to Yes in NetSapiens will appear as a transfer destination.The transfer instructions are pre-generated in a format the agent can use (e.g., “If the caller wants to speak to John Smith, transfer the call to pbx:1234”).[Screenshot: Transfer Destinations step]
7

Finish

Click Finish to create the agent. A success confirmation will appear, and your new agent will be added to the agent list, ready to use.[Screenshot: Success confirmation]

​
Agent Row Actions

​
Integrations (Gear Icon)

The Integrations view shows available third-party integrations that can be connected to the agent. Click on any integration to configure it. [Screenshot: Integrations view] Available integrations:
  • Google Calendar — Connect your Google account to manage your calendar
  • Outlook Calendar — Connect your Outlook account to manage your calendar
  • ModMed — Connect to ModMed EHR for patient scheduling and management
  • Model Context Protocol — Connect to external MCP servers for extended capabilities (extensions.mcp_servers; GET shape vs POST/PUT request shape)
  • Netsapiens — Connect to Netsapiens UCaaS platform for subscriber management
  • Twilio SMS — Connect the organization’s Twilio account (shared by all of its agents) and set this agent’s From Number. See Twilio

​
Usage Analytics (Magnifying Glass Icon)

Opens the per-agent Usage Analytics page, which displays:
  • Voice Minutes — A chart showing minute consumption over time
  • Call count — Total number of calls handled
  • Average duration — Average call length
[Screenshot: Usage Analytics view for a single agent]

​
Test Agent (Phone Icon)

Opens a built-in WebRTC test client that allows you to have a live voice conversation with the agent directly from your browser. Click Start Conversation to begin speaking with the agent and test its behavior. [Screenshot: Test agent WebRTC client]

​
Outbound agents

For an outbound agent the page also shows Test an outbound call, beside the browser test (below it on a narrow screen, and hidden while a browser conversation is running). Only users who can edit agents (Office Manager and above) see this card. Enter a phone number in international format (for example +15105550123) and, optionally, a greeting, a voicemail message and call context, then click Place call. The agent calls that number. The voicemail message is only used when the agent has Answering Machine Detection on. A test call is a real outbound call, so the same rules apply as to any other: the suppression list, calling hours, the agent’s contact limits and retry policy, and its concurrent-call limit. When a call is refused, the reason is shown inside the card. A call placed outside the agent’s calling hours is refused, unless the agent’s retry policy has Reschedule out-of-window calls on: then it is held and dials when the next window opens. If nobody answers, the agent’s default retry policy can call again later, and each retry is a real call too. An agent that requires a consent basis on every call can’t be tested this way, because the test form has no consent field. The Test calls list under the form shows the 20 most recent test calls placed from this page for this agent, by anyone in your organization, newest first, with each call’s attempt number and status. Calls still in progress update by themselves. A call waiting for its time, either a scheduled retry or a call held for calling hours, shows when it will dial and a Cancel button. Calls placed through the API or from the Dashboard don’t appear in this list; every call, test calls included, is listed in the Dashboard under Reports → Outbound Calls. [Screenshot: Test an outbound call card]

​
Edit Agent (Pencil Icon)

Opens the agent editor. See the Editing an Agent section below for details.

​
Delete Agent (Red X Icon)

Removes the agent. You will be prompted to confirm the deletion.
Deleting an agent is permanent and cannot be undone. All agent configuration, call history, and associated data will be removed.

​
Editing an Agent

The agent editor is organized into tabs. The available tabs depend on whether the agent is attached to a template or has been detached. When attached to a template: Basic Settings, SIP Settings, Rename Agent, Embeds When detached from a template: Basic Settings, Advanced Settings, SIP Settings, Rename Agent, Embeds

​
Attached vs. Detached Mode

When an agent is created from a template, it remains attached to that template. In attached mode, the editor shows a simplified set of fields tailored to the template type. A Detach button appears in the top-right corner of the editor. Clicking Detach permanently separates the agent from its template, unlocking the full set of editing options including the Advanced Settings tab. This gives you direct access to agent instructions, transfer rules, LLM model selection, and more.
Detaching an agent from its template cannot be undone. Once detached, the agent will no longer receive updates from the template.
[Screenshot: Detach confirmation dialog]

​
Basic Settings (Attached to Template)

When attached to a template, the Basic Settings tab shows template-specific fields:
  • Template — The parent template (e.g., Receptionist Agent) as a dropdown
  • Language — Speech-to-Text language (English - US, Spanish - US, or Multilingual)
  • Time Zone — The agent’s time zone for scheduling and time-related responses
  • Voice — Select a TTS voice from the dropdown, with a Preview button to hear a sample
  • Voice Speed — A slider to adjust speaking speed from Slower to Faster (default 1.0x)
  • Business name — How the agent refers to your business
  • Business website — Your company website URL
  • Additional Link — An extra URL for the agent’s knowledge base
  • Agent Notification Email — One or more email addresses for notifications (use Add Email to add additional recipients)
  • Greeting — The message your agent says when answering calls
  • Enable Transfers — Checkbox to enable call transfer functionality
  • Transfer Destinations — Auto-populated transfer instructions from the NetSapiens directory (with a refresh button to re-sync)
  • Agent instructions — What information the agent should know (with AI Prompt Improver button)
  • Questions and Info Collection — Define what information the agent should gather from callers
[Screenshot: Basic Settings - attached to template]

​
Basic Settings (Detached from Template)

When detached, the Basic Settings tab shows a different layout with these fields:
  • Greeting — The message your agent says when answering calls
  • Agent Notification Email — One or more email addresses (use Add Email to add more)
  • Links — Knowledge base URLs for the agent (use + Add to add links)
  • Functions — Agent functions like webhooks and email (use + Add to add functions)
  • Campaign ID — A unique identifier for the agent (with Copy button)
  • Language — Speech-to-Text language selection
  • Time Zone — The agent’s time zone
  • Voice — TTS voice selection with Preview button
  • Voice Speed — Slider from Slower to Faster (default 1.0x)
[Screenshot: Basic Settings - detached from template]

​
Advanced Settings (Detached Only)

The Advanced Settings tab is only available after detaching from a template. It provides full control over the agent’s core configuration:
  • Agent Instructions — The complete instructions prompt with AI Prompt Improver button to help refine your prompt
  • Transfer Rules — Instructions for when and how to transfer calls, with a refresh button to re-sync destinations from the NetSapiens directory
  • Model — The LLM model powering the agent’s intelligence (e.g., GPT-4.1 Smart)
  • STT Engine — The Speech-to-Text engine for transcribing caller speech (e.g., Deepgram)
  • Call Recording — Toggle to enable or disable call recording
  • Mobile Call Detection — Toggle to detect whether callers are using a mobile phone before the call starts
  • Background Audio — Toggle to enable ambient background sounds during calls
  • Screenshare — Allow Screenshare from Web Embedded Agent (LiveKit agents only): lets the agent ask a caller on the web widget to share a browser tab, window, or screen
  • Allow callers to interrupt the AI greeting — lets a caller talk over the AI-personalized greeting
  • Keypad (DTMF) Transfers — lets a caller reach a transfer destination by keying it, so a rule that transfers to pbx:100 also fires when they press 100
  • Hand-off Instructions — under Warm Transfer: what the AI says to the human on the consult leg, instead of the built-in briefing. See Warm Transfer
  • Filler While Waiting — speak a short filler line when a reply or tool is slow, with a configurable delay and your own phrases
For outbound agents the tab also offers:
  • Default Retry Policy — the retry policy the agent’s calls follow
  • Answering Machine Detection — work out who picked up: a person, a voicemail box, or an IVR menu, and what to do in each case. See Answering Machine Detection
  • IVR Menu Navigation — let the agent work through a phone menu to a person or department. Requires Answering Machine Detection. See IVR Menu Navigation
The Voice picker groups voices by the agent’s language — Spanish voices first for a Spanish agent, say — and labels each voice’s language; a note under Language explains that the accent comes from the voice. Preview plays each voice in its own language. [Screenshot: Advanced Settings tab]
Screenshare also needs the organization’s Screenshare policy switched on, and that is configured in the Dashboard only — Connect has no organization-level controls for it. Retained frames are viewable in the Dashboard as well; Connect’s call logs show the share record (consent, surface, who started it, start and end, why it ended) but not the frames. See Screenshare.

​
Adding Functions

Click the + Add button next to Functions in the Basic Settings tab (detached mode) to add capabilities to your agent. The following functions are available: [Screenshot: Add Functions dialog]
Send HTTP requests to your backend systems during or around calls. Use webhooks to fetch dynamic information, push data to a CRM, trigger workflows, or log call events.When adding a webhook, you configure the following fields:
  • Name (required) — Internal name for the webhook with no spaces (e.g., send_lead_to_crm)
  • Friendly Name — Display name shown in the UI (e.g., “Send Lead to CRM”)
  • Description — Describe when the agent should trigger this webhook (e.g., “Trigger when a caller provides their contact information and wants to be contacted later”)
  • Trigger Condition (required) — When the webhook fires. Options: Before Call, During Call, or After Call. Limits: 1 Before Call, 2 During Call, 1 After Call.
  • HTTP Method (required) — The HTTP method to use (e.g., POST, GET)
  • URL (required) — The endpoint URL to send the request to
  • Headers — Custom HTTP headers to include with the request (use + Add Header to add key-value pairs)
  • Parameters — Custom parameters to include in the request body (use + Add Parameter to add key-value pairs)
[Screenshot: Configure Webhook form]

​
Headers Example

Headers are useful for authentication and specifying content types. Each header is a key-value pair. Common examples:

​
Parameters Example

Parameters define the data sent in the request body. Each parameter is a key-value pair that the agent will populate based on the call context. Common examples:
Parameter values act as descriptions that tell the agent what data to extract from the conversation and include in the webhook request.
When enabled, your agent can send emails during the call based on the instructions you provide. This is useful for sending confirmations, follow-ups, or notifications to team members in real time while the caller is still on the line.
Transfer the call to another AI agent based on conditions you specify. This allows you to build multi-agent workflows where different agents handle different types of inquiries.
Click the + Add button next to Links in the Basic Settings tab to provide your agent with external knowledge sources. Links point to web pages or PDFs that the agent can reference during calls to provide accurate, up-to-date information.
Links serve as the agent’s knowledge base. The agent will use the content from these URLs to answer caller questions. Link to specific, relevant pages such as FAQ sections or product detail pages rather than general homepages.

​
Sharing and Publishing

Once your agent is configured, you can make it available to receive calls and interact with customers.

​
Forwarding Calls

Forward your existing business phone number to your agent’s Telzino number so that incoming calls are answered by the AI agent.

​
Publishing a Profile

Publish a public profile page for your agent that allows people to interact with it via web chat. The profile page can be shared via a link.

​
Embedding on a Website

Embed your agent as a chat widget on your website so visitors can interact with it directly from your web pages.

​
Porting a Phone Number

Port an existing phone number to Telzino so your agent answers calls on your current business number.