Skip to main content

Creating a Reply Assistant for Your Campaigns

This article explains how to create a Reply Assistant in Alta and attach it to your campaigns, so it can help manage and classify prospect replies based on your selected use case.

Written by Katie Supporté

What Is a Reply Assistant?

A Reply Assistant is an AI assistant that helps respond to replies from prospects on LinkedIn or email.

You can configure the assistant based on your workflow, define its goal, add relevant knowledge, select tools, choose a tone of voice, and define possible outcomes.

For example, you can create a Reply Assistant that qualifies leads using the BANT framework.

BANT stands for:

  • Budget

  • Authority

  • Need

  • Timeline


How to Create a Reply Assistant

Step 1: Go to Assistants

From the left-hand side menu in Alta, click Assistants.


Step 2: Click Create Assistant

Click Create Assistant.


Step 3: Select Reply Assistant

Choose Reply Assistant.

This will allow you to create an assistant that can handle replies from prospects.


Step 4: Choose a Use Case

You can either:

  • Start from scratch

  • Select a predefined use case

For this example, select Lead Qualification.

The setup process is similar for each use case.


Setting Up the Reply Assistant

Step 5: Define the Assistant Goal

Start by defining the goal of the assistant.

For the Lead Qualification use case, the goal is to qualify leads using the BANT framework.

This means the assistant will review the prospect’s reply and determine whether they meet the qualification criteria.

You can also tell the assistant not to reply in a specific situation. In the goal, name that situation and say not to reply. The assistant skips that message and does not send one.


Step 6: Add a Knowledge Base

Next, you can add a Knowledge Base.

The Knowledge Base helps the assistant respond with accurate and relevant information.

Use this section to provide helpful context such as:

  • Company information

  • Product details

  • Pricing information

  • Qualification criteria

  • Common objections

  • FAQs

  • Messaging guidelines

This gives the assistant the information it needs to reply more effectively.


Configuring Outcomes

After the goal, open Output and configure the possible outcomes.

Outcomes help the assistant classify replies based on the prospect’s response.

When a conversation matches an outcome, the assistant can send a closing message. Each outcome has an optional Closing message prompt for that message, on every channel. Write that no message should be sent to skip the closing for that outcome. To skip the closing on every outcome, say the same in the goal.

Step 7: Go to Output

Click Output.


Step 8: Add Outcome: Qualified

Create an outcome called Qualified.

Use this outcome when the prospect confirms all four BANT criteria:

  • Budget

  • Authority

  • Need

  • Timeline

Example outcome prompt:

The prospect confirms all four BANT criteria: Budget, Authority, Need, and Timeline.

You can also select a color for this status.


Step 9: Add Outcome: Unqualified

Create an outcome called Unqualified.

Use this outcome when the prospect is missing multiple BANT criteria.

Example outcome prompt:

The prospect is missing multiple BANT criteria and is not currently a qualified lead.

You can also select a color for this status.


Step 10: Add Outcome: Needs Nurturing

Create an outcome called Needs Nurturing.

Use this outcome when the prospect meets some, but not all, of the BANT criteria.

Example outcome prompt:

The prospect has met some BANT criteria but is not fully qualified yet and may need further nurturing.

You can also select a color for this status.


Step 11: Go to Configurations

Click Configurations.

Model, tools, tone, response length, attachments, and stops are on this step.


Step 12: Add Attachments

In the Attachments section, you can choose files from your Knowledge Base that the assistant may attach to a reply — for example a one-pager, a pricing sheet, or a case study.

The assistant attaches a file only when the prospect asks for one, or when your goal instructions describe a situation that calls for it.

Attachments are sent on email replies only. On LinkedIn, SMS, and WhatsApp the assistant replies without the file, and it will not say that a file is attached. If a prospect asks for a document on one of those channels, the assistant answers in the message itself or offers to send the file by email, and your team can send it manually.


Step 13: Guide the Assistant’s Tools

Next, guide the assistant on which tools it can use to complete the task.

This helps the assistant understand what actions it can take while managing replies.

For example, depending on your setup, the assistant may be able to:

  • Review prospect information

  • Use campaign context

  • Reference knowledge base content

  • Help classify replies

  • Support follow-up actions


Step 14: Choose the Tone of Voice

Select the assistant’s tone of voice.

Choose the tone that best matches your brand and the way you want the assistant to communicate with prospects.


Step 15: Choose the Response Length

Select the assistant’s response length.

This helps keep replies clear, complete, and not too long.


On Configurations, When the reply agent stops lists the stops. Two of them can be turned off: an outcome is found, and someone on your team messaged after an auto-reply. The other stops stay on. When a conversation matches an outcome and that stop is on, the assistant saves the outcome, sends the closing message, and stops. Later replies also stop, because the prospect already has an outcome. If you turn that stop off, the assistant still saves the outcome and sends a normal reply instead of the closing message, and later replies continue.

Step 16: Save the Reply Assistant

Once you are happy with the setup, click Save.

Your Reply Assistant is now ready to use.


Attaching the Reply Assistant to a Campaign

After saving the Reply Assistant, you can attach it to a campaign.

This allows the assistant to manage or support replies for prospects in that campaign.

Step 1: Open the Relevant Campaign

Go to the campaign where you want to use the Reply Assistant.


Step 2: Attach the Reply Assistant

Select the Reply Assistant you created and attach it to the campaign.

Once attached, the assistant will use the setup, knowledge base, tone, response length, and outcomes you configured.


Example Outcome Setup

Here is an example setup for a Lead Qualification Reply Assistant:

Outcome

When to Use It

Qualified

The prospect confirms Budget, Authority, Need, and Timeline

Unqualified

The prospect is missing multiple BANT criteria

Needs Nurturing

The prospect meets some BANT criteria but not all


Best Practices

Use a Clear Goal

Make sure the assistant’s goal is specific.

For example, instead of saying “respond to leads,” use a goal like:

Qualify leads using the BANT framework and classify them based on their replies.


Add Relevant Knowledge Base Content

The better your Knowledge Base is, the better the assistant’s replies will be.

Include information the assistant may need to answer common questions or qualify prospects properly.


Keep the Tone Consistent

Choose a tone that matches your company’s communication style.

For most sales workflows, Professional or Friendly works best.


Use Standard Response Length

Standard response length is usually the best option because it keeps replies helpful without making them too long.


Define Outcomes Clearly

Each outcome should have a clear description.

This helps the assistant classify replies correctly and makes it easier for your team to understand the prospect’s status.


Summary

A Reply Assistant helps manage LinkedIn and email replies in Alta.

To create one, go to Assistants → Create Assistant → Reply Assistant, choose a use case such as Lead Qualification, define the assistant’s goal, add a Knowledge Base, define outcomes, then configure tools, tone, and response length.

Once saved, attach the Reply Assistant to a campaign so it can support reply handling and lead qualification.

Did this answer your question?