Create a Reply Agent
In this guide, you'll learn how to create a Reply Agent, connect an AI provider, configure channels and labels, customize the agent's behavior, enable follow-ups, connect your calendar, and provide knowledge so your agent can generate accurate, context-aware replies.
What is a Reply Agent?
A Reply Agent is an AI-powered assistant that helps you manage conversations across your connected communication channels.
Depending on how it's configured, a Reply Agent can:
Draft replies for review
Automatically send replies
Generate follow-up messages
Respond using your company knowledge
Suggest available meeting times through calendar integrations
Personalize responses through custom prompts
Reply Agents work across your selected channels and only respond to conversations that match the labels and settings you define.
Why Use a Reply Agent?
Reply Agents help teams respond faster while maintaining consistent, high-quality communication.
Benefits include:
Faster response times
Consistent messaging
Reduced manual work
Automated follow-ups
AI-assisted meeting scheduling
Better scalability for high-volume inboxes
Step 1: Create a New Reply Agent
Go to:
Settings → Reply Agents
Click Create Agent.
Step 2: Configure Basic Settings
Agent Name
Give your Reply Agent a descriptive name.
Examples:
Sales Reply Agent
Support Agent
LinkedIn Agent
Demo Booking Agent
John Wick
Agent Mode
Choose how your Reply Agent should operate.
Human in the Loop
The Reply Agent drafts replies for review before they are sent.
This mode is recommended when you want a team member to approve every response.
Fully Automated
The Reply Agent generates and sends replies automatically without manual approval.
This is useful for repetitive workflows such as:
Not Interested responses
Frequently asked questions
Automated follow-ups
Routine inbound replies
Step 3: Configure Your AI Provider
Choose the AI provider your Reply Agent will use.
Supported providers include:
OpenAI
Anthropic
OpenRouter
vLLM
Depending on your selected provider, you'll be asked to configure the required credentials.
For example:
OpenAI API Key
Anthropic API Key
OpenRouter API Key
vLLM Base URL, Model Name, and API Key (if required)
Test Your Connection
Click Test to verify your AI provider credentials.
If the connection is successful, MasterInbox will confirm that your API key has been validated.
If validation fails, verify your credentials before continuing.
Step 4: Select Channels
Choose which connected channels this Reply Agent should monitor.
You can assign the agent to:
Google accounts
Outlook accounts
LinkedIn accounts
Integrated tools
The Reply Agent will only generate replies for conversations received through the selected channels.
Example
Create one Reply Agent for email conversations and another specifically for LinkedIn messages.
Step 5: Select Labels
Choose which labels should trigger the Reply Agent.
Examples include:
Interested
Information Request
Meeting Request
Not Interested
Limiting Reply Agents to relevant labels helps improve reply quality and reduces unnecessary AI usage.
You can also create specialized Reply Agents for different stages of your pipeline.
Step 6: Configure Agent Behavior
Customize how your Reply Agent writes replies.
Available settings include:
Tone of Voice
Response Length
AI Model
Creativity Level (Temperature)
You can enable capabilities such as:
Handle Follow-ups
Auto Respond to New Conversations
Stop Replies After Meeting is Booked
Tone of Voice
Choose the communication style your Reply Agent should use.
Available options:
Professional – Clear and business-focused.
Friendly – Warm and conversational.
Casual – Relaxed and informal.
Formal – Polished and professional.
Persuasive – Encourages prospects to take action.
Empathetic – Understanding and supportive.
Choose the tone that best matches your brand and audience.
Response Length
Choose how detailed AI-generated replies should be.
Available options:
Short – 1–2 sentences.
Medium – Approximately one paragraph.
Long – 2–3 paragraphs.
Variable – The Reply Agent automatically adjusts the response length based on the conversation.
These settings apply to all replies generated by the Reply Agent.
Select Campaigns
Choose which campaigns the Reply Agent should monitor.
By default, All Campaigns are selected, allowing the Reply Agent to respond to conversations regardless of which campaign or sequence the prospect came from.
If needed, you can limit the Reply Agent to specific campaigns by selecting only the campaigns you want it to monitor.
Recommended use cases
Create dedicated Reply Agents for different campaigns.
Use different prompts or reply styles for separate outbound campaigns.
Prevent a Reply Agent from responding to conversations outside a specific campaign.
Note: Only campaigns synchronized with your connected outreach platform are available for selection. If All Campaigns is selected, the Reply Agent will process conversations from every synchronized campaign.
Handle Follow-ups
When enabled, the Reply Agent can automatically send follow-up messages if a prospect doesn't reply.
You can create as many follow-up steps as needed, with each step having its own delay and message template.
learn more:
Configure Follow-up Steps
For each follow-up step, you can configure:
Delay – Choose how many days to wait before sending the follow-up after the previous message.
Follow-up Template – Write the message the Reply Agent should send for that step.
You can also add or remove follow-up steps at any time.
Template Variables
Follow-up templates support variables to automatically personalize messages.
Available variables include:
Prospect Name
Prospect First Name
Prospect Email
Sender Name
Sender First Name
Sender Last Name
These variables are automatically replaced with the appropriate values when the follow-up is sent.
Learn more:
Common Use Cases
Build a multi-step follow-up sequence.
Keep outreach consistent without manually sending follow-ups.
Auto-Respond to New Conversations
When enabled, the Reply Agent automatically generates replies for new incoming conversations that match the agent's configured channels and labels.
This helps ensure new conversations receive a timely response without requiring manual intervention.
Common Use Cases
Automatically acknowledge new inquiries.
Reduce response times for new conversations.
Stop Replies After Meeting Is Booked
When enabled, the Reply Agent automatically stops generating replies once a meeting has been booked or the generated reply indicates that a meeting has been scheduled.
This helps prevent unnecessary follow-ups after the conversation has successfully reached the booking stage.
If needed, this behavior can be overridden for individual conversation threads.
Common Use Cases
Prevent duplicate follow-ups after scheduling a meeting.
End automated outreach once the objective has been achieved.
Keep conversations clean after a successful booking.
Step 7: Connect a Calendar
Reply Agents can use your connected calendar to suggest real meeting availability.
Supported calendar providers include:
Calendly
HubSpot Calendar
GoHighLevel Calendar
Before selecting a calendar, make sure it has already been connected under Settings → Integrations.
AI Model & Creativity Level
Choose an AI Model
Select the AI model your Reply Agent will use to generate replies.
The available models depend on the AI provider you've configured (OpenAI, Anthropic, OpenRouter, or vLLM).
Choose a model that best fits your requirements for speed, cost, and response quality.
Recommended: GPT-5 for OpenAI and Claude Sonnet 4.6 for Anthropic.
Creativity Level
The Creativity Level (Temperature) controls how creative or consistent the Reply Agent's responses are.
Lower values produce more focused, predictable, and consistent replies.
Higher values produce more creative and varied responses.
For most business communication, a lower to medium creativity level is recommended.
Note: Temperature is not applied to GPT-5 reasoning models, as they use a fixed internal setting.
Step 8: Configure Your Prompt & Agent Knowledge
Customize how your Reply Agent behaves using a Custom Prompt.
You can define:
Writing style
Business rules
Qualification criteria
Meeting instructions
Reply guidelines
And more
You can also provide agent knowledge by:
Uploading PDF, DOCX, XLSX, CSV, HTML, or TXT files
Pasting knowledge directly into the editor
Reply agents can use up to five uploaded documents as references when generating replies.
Learn more:
Step 9: Save Your Reply Agent
Review your settings and click Create Agent.
Your Reply Agent is now ready to begin handling conversations according to the configuration you've defined.
Clone Reply Agent
The Clone Reply Agent option creates an exact copy of an existing Reply Agent within the same workspace.
This is useful when you want to reuse the same configuration while changing only a few settings, such as different channels, labels, or prompts.
Common use cases:
Use the same configuration for different labels (e.g., Interested vs. Not Interested).
Create separate agents for Email and LinkedIn while keeping the same settings.
Quickly build variations of an existing Reply Agent without starting from scratch.
Duplicate Reply Agents Across Workspaces
When duplicating a workspace, you can choose to include your existing Reply Agents.
This copies the Reply Agents and their configurations into the new workspace, helping you get started faster without recreating each agent manually.
Note: After duplicating a workspace, review the copied Reply Agents to ensure their channels, integrations, and API credentials are configured correctly for the new workspace.
When duplicating a workspace, you can choose to include your existing Reply Agents.
This copies the Reply Agents and their configurations into the new workspace, helping you get started faster without recreating each agent manually.
Note: After duplicating a workspace, review the copied Reply Agents to ensure their channels, integrations, and API credentials are configured correctly for the new workspace.
Once your Reply Agent is created:
It monitors the selected channels.
It responds only to conversations that match your selected labels.
Replies are generated according to your configured AI provider, prompt, and behavior settings.
Follow-ups and meeting scheduling work automatically when enabled.
Company knowledge is used to generate more accurate and relevant replies.
Your Reply Agent is now ready to assist your team with faster, more consistent conversations.