Create with AI

Don’t want to start from scratch? Describe what you need in plain English, and our AI will generate a complete agent — whether Single Prompt or Conversational Flow.

[IMAGE: Full Create with AI page showing configuration panel on left and prompts on right]

How It Works

Choose Agent Type

Select whether you want a Single Prompt or Conversational Flow agent.

Configure Settings

Set voice, model, call direction, and optional knowledge base.

Write Four Prompts

Describe your agent’s role, conversation flow, behaviors, and end conditions.

Click Create

AI generates your complete agent.

[IMAGE: Loading modal showing “Analyzing requirements → Designing prompt → Creating your agent”]

Review & Refine

The editor opens with your generated agent. Customize as needed.

Pro tip from the platform: “Use caller context (like past tickets) to greet by name. Even a small touch of personalization boosts trust.”

Configuration Options

[IMAGE: Left panel showing all dropdowns and toggles]

The left panel controls your agent’s foundation:

Agent Type

[IMAGE: Dropdown showing Single Prompt and Conversational Flow options]

Type	What AI Generates
Single Prompt	One comprehensive prompt
Conversational Flow	Complete workflow with nodes and branches

Call Direction

[IMAGE: Dropdown showing Inbound and Outbound options]

Direction	Use Case
Inbound	Agent receives calls (support, inquiries)
Outbound	Agent makes calls (campaigns, reminders)

Emotive Model

[IMAGE: Toggle switch for emotive model]

Beta Feature: Emotive models use speech-to-speech technology for more natural, emotionally-aware responses. Experimental but impressive.

Voice Selection

[IMAGE: Voice selection dropdown with preview button]

Choose the voice your agent will use. Click ▶️ to preview before selecting.

Knowledge Base

[IMAGE: KB dropdown showing None and available KBs]

Optionally attach an existing knowledge base for the agent to reference.

The Four Prompts

[IMAGE: Right side with four text areas stacked]

Each prompt shapes a different aspect of your agent. All require minimum 50 characters.

1. Role & Objective
2. Conversational Flow
3. Dos, Don'ts & Fallbacks
4. End Conditions

[IMAGE: First prompt field highlighted]

What it defines: Who the agent is and what it’s trying to accomplish.Label: “Provide the role & objective of your agent”Good example:

You are a friendly and confident customer support agent for 
a fintech platform. Your goal is to quickly understand the 
user's issue, respond with empathy, and provide clear, 
step-by-step solutions that make them feel heard and supported.

Maintain a calm and reassuring tone throughout the conversation, 
ensuring that the user feels confident in your assistance. Only 
mark the issue as resolved once the user has confirmed they are 
fully satisfied with the outcome.

Tips:

Be specific about personality (friendly, professional, casual)
State the primary goal clearly
Mention the industry/context

[IMAGE: Second prompt field highlighted]

What it defines: The step-by-step conversation structure.Label: “Provide conversational flow that the agent should follow”Good example:

As a NorthLoop Support agent, greet customers warmly and 
confirm they are NorthLoop customers. Ask one focused question 
to identify if the issue is billing or connectivity.

For connectivity, get the service address; for billing, collect 
the email or phone number. Verify the account and, if needed, 
try an alternate identifier.

Diagnose the issue: check for outages first, then guide through 
reboot or cabling steps. If the outage lasted over four hours, 
issue a refund. If local fixes fail, schedule a technician.

Confirm resolution and next steps, invite questions, and close 
politely.

Tips:

Describe the conversation as a journey
Include decision points (if X, then Y)
Cover the main paths callers will take

[IMAGE: Third prompt field highlighted]

What it defines: Behavioral guardrails and edge case handling.Label: “Set your agent’s dos, don’ts, and fallback behaviours — especially for edge cases or sensitive situations”Good example:

Your agent should always remain calm, polite, and solution-oriented.

DO:
- Listen actively and show empathy
- Provide clear next steps at every stage
- Confirm understanding before proceeding
- Escalate politely when needed

DON'T:
- Make assumptions about the issue
- Overpromise on timelines or resolutions
- Share unverified information
- Rush the customer

FALLBACKS:
- Account lookup failures → Offer alternate verification methods
- Incomplete verification → Politely re-request information
- Unclear issues → Ask clarifying questions, don't guess
- System limitations → Acknowledge honestly and offer alternatives

Always prioritize accuracy, empathy, and clarity over speed.

Tips:

List specific behaviors, not vague goals
Cover sensitive scenarios (angry callers, edge cases)
Define what to do when things go wrong

[IMAGE: Fourth prompt field highlighted]

What it defines: When and how the conversation should end.Label: “Specify the conditions under which the call should end”Good example:

End the call when:

1. Issue fully resolved and customer confirms satisfaction
   - Thank them for calling
   - Offer additional help before closing

2. Customer requests to end the call
   - Acknowledge and close gracefully
   - Offer callback if issue unresolved

3. Successful transfer to human agent
   - Confirm transfer is happening
   - Brief goodbye

4. Customer becomes unresponsive
   - After 3 attempts to re-engage
   - Leave callback information

5. Issue outside scope
   - Acknowledge limitation
   - Provide alternative resources
   - End with helpful next steps

Tips:

Cover all the ways a call might end
Include what to say in each scenario
Don’t forget edge cases (no response, wrong department)

Using Templates

[IMAGE: Row of template tabs: Real Estate, Credit Card, Customer Support, etc.]

Templates pre-fill all four prompts with tested content for common use cases.

Available Templates

Template	Best For
Real Estate	Property inquiries, viewing scheduling
Credit Card	Card services, payment questions
Customer Support Electronics	Tech support, troubleshooting
General Sales	Lead qualification, product info
Onboarding Assist	New user guidance, setup help

How to Use

Click a Template Tab

[IMAGE: Template tab being selected]

Prompts Auto-Fill

All four prompt fields populate with template content.

[IMAGE: All four fields now containing text]

Customize

Update company names, policies, and specifics for your business.

Create

Click Create agent to generate.

Clear All

[IMAGE: Clear All button location]

Click Clear All to reset all prompts and start fresh.

Tips for Single Prompt Agents

When generating Single Prompt agents:

Focus on personality and knowledge

Since there’s no visual flow, the prompt needs to cover everything the agent should know and how it should behave.Include:

Detailed role description
All topics it should handle
Communication style
Edge case handling

Be comprehensive in conversation flow

Even without visual nodes, describe the typical conversation arc:

Start with a warm greeting. Ask how you can help. 
Listen to the issue. Acknowledge and confirm understanding.
Provide the solution or collect needed information.
Confirm resolution. Offer additional help. Close warmly.

Define flexibility

Single Prompt agents should handle topic jumps gracefully:

If the customer changes topics, acknowledge the shift and 
adapt. Don't force them back to previous subjects.

Tips for Conversational Flow Agents

When generating Conversational Flow agents:

Describe explicit branches

Tell the AI exactly when paths should split:

After greeting, ask if calling about billing or technical.

BILLING PATH:
- Verify account
- Ask about specific charge
- Explain or process refund

TECHNICAL PATH:
- Identify product
- Troubleshoot steps
- Schedule technician if needed

Specify data collection

List what information must be gathered and when:

Must collect before qualifying:
Company name
Company size (employees)
Current solution
Budget range
Decision timeline

Define all endings

Every path needs a conclusion:

End points:
- Qualified lead → Transfer to sales with context
- Not qualified → Send resources, end call
- Requested callback → Schedule and confirm
- Angry customer → Escalate to manager

After Generation

Once AI creates your agent:

[IMAGE: Editor with AI-generated content loaded]

Review the Output

Read through the generated prompt (Single Prompt) or walk through the flow (Conversational Flow).

Test Immediately

Click Test Agent to try a conversation before making changes.

Refine

Adjust anything that doesn’t match your vision:

Edit prompts
Rearrange nodes
Add missing branches
Update conditions

Configure Settings

Add features from the Configuration Panel:

End Call triggers
Transfer numbers
Knowledge Base
Variables
API Calls

Next Steps

Single Prompt Setup

Refine your Single Prompt agent

Conversational Flow Setup

Master workflow building

Configure Your Agent

Voice, model, and features

Test Your Agent

Validate before deploying

Introduction

Single Prompt Agents

Conversational Flow Agents

Features

Deployment

Analytics & Logs

Cookbooks

Troubleshooting

How It Works

Configuration Options

Agent Type

Call Direction

Emotive Model

Voice Selection

Knowledge Base

The Four Prompts

Using Templates

Available Templates

How to Use

Clear All

Tips for Single Prompt Agents

Tips for Conversational Flow Agents

After Generation

Next Steps

Single Prompt Setup

Conversational Flow Setup

Configure Your Agent

Test Your Agent

Introduction

Single Prompt Agents

Conversational Flow Agents

Features

Deployment

Analytics & Logs

Cookbooks

Troubleshooting

​How It Works

​Configuration Options

​Agent Type

​Call Direction

​Emotive Model

​Voice Selection

​Knowledge Base

​The Four Prompts

​Using Templates

​Available Templates

​How to Use

​Clear All

​Tips for Single Prompt Agents

​Tips for Conversational Flow Agents

​After Generation

​Next Steps

Single Prompt Setup

Conversational Flow Setup

Configure Your Agent

Test Your Agent

How It Works

Configuration Options

Agent Type

Call Direction

Emotive Model

Voice Selection

Knowledge Base

The Four Prompts

Using Templates

Available Templates

How to Use

Clear All

Tips for Single Prompt Agents

Tips for Conversational Flow Agents

After Generation

Next Steps