Skip to main content

Creating an AI Agent

Yeshawanth Reddy
Updated

This article explains how to create an AI agent using the AI Agent Builder. Creating an agent involves defining what the agent should do, selecting the blocks it can use, enabling channels, and publishing the agent so it can interact with candidates.

Prerequisites

To set up your AI agent, ensure you complete the following prerequisite steps:

  • Secure Admin Access: You must have administrative privileges within Sense.
  • Verify ATS Connection: Confirm that your Applicant Tracking System (ATS) is properly connected and that all necessary field mappings have been established.
  • Define Use Case: Clearly determine the purpose of the agent (e.g., screening candidates, sourcing new talent, or data collection).

Creating an AI Agent

Creating an AI Agent allows you to build an automated assistant that can interact with candidates and perform recruiting tasks based on a defined use case.

Each Agent is created to achieve a specific goal, such as:

  • Application screening
  • Talent screening
  • Talent sourcing
  • Data enrichment

What the AI Agent is Built With

An AI Agent is made up of modules (blocks), where each module represents a specific capability.

For example:

  • Greeting – starts the conversation
  • Data Collection – captures candidate information
  • Screening – evaluates candidates
  • FAQ – answers candidate questions
  • Closing & Feedback – ends the conversation

These modules work together to complete the Agent’s goal.

Multi-Channel Capability

The AI Agent supports seamless communication with candidates across multiple channels, including:

  • Chat (Web)
  • Chat (SMS)
  • Voice

Understand the Default Modules

The agent is built using modules, and you’ll work with the following modules and refer to the images for your reference:

  • Greeting — starts the conversation

  • Data Collection — captures candidate details

  • Screening — asks role-specific questions (This module is available for the outbound use cases, i.e. for Application Screening & Talent Screening)

  • FAQ — answers candidate questions

  • Closing & Feedback — ends the conversation by collecting feedback (if enabled)

Note: These default modules are automatically created and are required for the Agent to function correctly and can be disabled but cannot be deleted, but you can configure them as per your requirements. Some modules, such as Greeting and Closing & Feedback, are always constant by default, while others vary depending on the use case, such as screening, sourcing, etc.

Create a New Agent - Outbound (Screening & Data Enrichment)

  1. Login to your Sense Dashboard.
  2. Navigate to the AI Agents (Agent Builder) from the side menu.
  3. Give a brief description of what to create in the description section and click Create Agent.
  4. Alternatively, you can get started with the existing templates from the existing ones below.

Once submitted, the system initiates drafting an agent with default modules based on your use case or selection. The Modules in the creation section are discussed in the upcoming sections.

Create a New Agent - Inbound

  1. Login to your Sense Dashboard.
  2. Navigate to the AI Agents (Agent Builder) from the side menu.
  3. Give a brief description of what to create in the description section, and click Create Agent.
  4. Alternatively, you can get started with the existing templates from the existing ones below.

Once submitted, the system initiates drafting an agent with default modules based on your use case or selection. The Modules in the creation section are discussed in the upcoming sections.

Module Configurations

To begin using the drafted AI Agent, you must open and configure each module according to your specific requirements.

Greeting

  1. Click on the “Greeting” module to edit or customize the greeting message.
  2. There will be a standard message by default, but you can personalize it as per your needs.
  3. You can use the variable picker and emojis for personalization of the greeting message.

Data Collection

  1. Click on the Data Collection or Configure to edit the module
  2. Add static questions to capture candidate details and get files such as resume, cover letter, reference letter, work authorization, etc as per your requirements. 
  3. Map responses to ATS fields (optional).
  4. Define instructions and conditions that you want the agent to follow (optional).
  5. Use Additional Configurations, if needed:
    1. Enable confirmation of collected data - Turn this on to show a summary of the information collected from the user and allow them to review and confirm it before proceeding.
    2. Trigger email notification - Turn this on to send an email notification with the collected user details after the data collection step is completed.

Note: Minimum 1 question is required and the added questions are auto saved. ATS fields and instructions are optional and can be used if needed only. 

Restriction: Do not use "File Upload" questions in case your required channel is Voice. File upload question types are not supported in the Voice channel.

FAQ

  1. Click on the FAQ to edit the module
  2. Select the Brand for which the AI Agent should answer FAQs. If there is only one Brand, the Brand is auto-selected and locked. 
  3. Under Content Access, choose the Knowledge Base content that the Agent can refer to while answering user questions. You can search and select specific content or keep full access enabled. 
    1. Full Access is selected by default, and you can change it if needed.
  4. Enable Trigger email notification if you want to be notified when the Agent is unable to answer a question and a fallback is triggered.
  5. In Recipient Email, select the email variable where the notification should be sent. (Optional) Use CC to add additional recipients who should receive the notification. (This can be used only for the candidate entity)
  6. Once all configurations are completed, click Save to apply the changes.

Closing & Feedback

  1. Click on the Closing to edit the module.
  2. Customize how the Agent ends the conversation. Update the message as needed to clearly inform the user about the next steps.
  3. Use the available options to insert variables, question responses or emojis, if required.
  4. Ensure the closing message is complete and user-friendly, as this will be the final communication shown to the user.
  5. Enable Feedback Collection to collect feedback at the end of the conversation.
  6. Once enabled, customize the feedback question that will be shown to the user.
  7. Click Save to apply the changes.

Note: You can use the variable and question response pickers to create a custom closing message, and add emojis if you like.

Edit a Published AI Agent

You can edit a published AI agent directly without duplicating or archiving it. Your changes remain unpublished until you re-publish the agent, so candidates can continue interacting with the current live version while you edit and test your updates.

You can update module content, enable or disable modules, add new modules, and modify channel settings. Once you re-publish, new conversations use the updated agent, while ongoing conversations continue on the version they started with.

For detailed instructions, refer to the Edit the Published AI Agent article.

Tip: After building the Agent, it is highly recommended to test it before publishing and actively using it. Please refer to the following article for detailed instructions on testing the built AI Agent: Testing an AI Agent

Related articles