Skip to content

Prompt Agents

As an administrator, you can create Prompt Agents in the prompt agent builder that users can utilize. These agents collect specific user inputs as variables to customize prompts, enabling various use cases such as content generation, language translation, and text summarization.

Prompt agents operate using two distinct components:

  • User-Collected Inputs: In this scenario, the user's inputs are incorporated into the prompt as variables to generate the required response. For instance, a job description (JD) can be created using user inputs regarding experience, responsibilities, and required skills.

  • Administrator-Uploaded Knowledge: In this scenario, an administrator can upload a file to be used as a template. For example, a JD can be generated based on user-provided inputs and a pre-uploaded template that the agent will utilize to ensure consistency and adherence to specific formatting guidelines.

You can create a new agent or import an existing one.

Import existing Prompt Agent

To import an existing Prompt Agent:

  1. Click the Import Agent button located in the upper-right corner.

  2. Select the .ZIP file of the existing agent.

  3. Click Import to complete the process. The imported agent will appear on the Agents page.

Create a new Prompt Agent

You can create a new prompt agent to assist users in completing various tasks by utilizing the prompt capabilities.

To create a new Prompt Agent, follow these steps:

  1. In the Admin Console, click AI Agents from the left pane, and then select Prompt Agents. A list of available prompt agents will be displayed.

  2. On the Prompt Agents page, click +Create Agent.

  3. The Prompt Agent creation wizard will take you through the following steps:

    1. Step 1: Details and Purpose

    2. Step 2: Source Configuration

    3. Step 3: Preview the Agent configuration

    4. Step 4: Publish the Agent

Step 1: Details and Purpose

Provide a suitable and unique name for the agent and describe its purpose. Defining the agent’s purpose enables the system to accurately recognize the agent’s capabilities and effectively utilize them to respond to user queries aligned with the specified intent. It is essential to clearly outline the specific use cases for which the agent is designed, as all the fields, queries, and prompts are auto-generated based on the purpose provided.

To change the agent icon, click on the agent logo located above the agent's name. You can choose from a list of predefined logos or upload a custom logo. Once you're ready, click Continue to proceed to the next step.

Step 2: Source

In the Sources section, you can configure the following fields

The parameters fields are provided based on the purpose of the agent you provided in the previous step. For example, if the purpose is "summarization", fields might include Word Count, Summary Format etc.

User Input

To configure user input, follow the steps outlined below:

  1. Click on the +Add context.

  2. Enter context name and any placeholder text in the placeholder text field if required. The placeholder text will appear in the form field before the end user provides input.

  3. Select the Field type from the list for each field. You can select a single-line, multiline, single-select, multi-select, number, file upload, or URL based on the fields.

  4. Click the mandatory toggle button ON if you want the field to be mandatory and click Done.

  5. In the parameter section, parameter fields are prepopulated based on the agent’s purpose. Enter parameter details and click on + Add field if required.

  6. Enter a name and any text in the placeholder text field and in Description, enter a detailed description of the field to ensure automated field population when the LLM processes natural language queries and matches entities to their corresponding fields.

    For example, if you need a field that captures organizational departments, rather just naming it as Division, include a description explaining that this field accepts internal department names, helping the LLM identify the department name from the user's query and pre-populate it in the Division field.

  7. Select the Field type from the list for each field. You can select a single-line, multiline, single-select, multi-select, number, file upload, or URL based on the fields.

  8. Toggle the Allow Upload Files or Allow URL Content option to ON to enable file uploads or retrieval from URLs. This option is available only for Single-line and Multiline text field types.

  9. Click the mandatory toggle button ON if you want the field to be mandatory and click Done.

Knowledge

  1. Click the + Upload button to upload a file containing knowledge that the Prompt agent can use to generate responses. This uploaded information may also be utilized in prompts. For example, if you uploaded a “Sample JD template” file, you can write a prompt that says, “Create a JD which is in the same format as in the Sample JD template file.”

    Note: If the file size exceeds the context size limitation, it impacts the models, and an error message is displayed along with a list of models. Only a limited amount of information is used as context for generation. However, when the file limit is exceeded, all uploaded knowledge files are referenced for answering but not for generation.

  2. Select the model required from the Model selection drop-down list, which displays all the integrated models. To know more about model integration, click here.

  3. Toggle the Show to users button to display this prompt to the user. After switching the toggle to ON, you can choose between Read-only or Editable modes.

Prompts

A default prompt is automatically generated based on the purpose defined earlier and can be customized using variables. For example, if the purpose is "Job Finder," the default prompt might include, “Company Name, Job Title, Responsibilities etc”

Enable the Show to users toggle button to show this prompt to the user. Once you click the toggle button to ON, you can select whether you want it in Read-only or Editable.

Multiprompt

For a multi-prompt setup within the agent builder, users can generate multiple prompts and select one from the prompt selector option field based on the specific context.

To add another prompt, click on the + Add another prompt button. You can rename prompts and use drag-and-drop to reorder them. Click on the Generate prompt button to generate a prompt based on your form’s purpose and field values.

Once you add multiple prompts, the prompt option will appear under the prompt selector.

  1. Click on the edit option or anywhere within the prompt area to expand it and view the available options

  2. For the prompt selector, you can edit field name and any placeholder text in the placeholder text field if required

    Note: The Mandatory toggle field is enabled by default and cannot be modified by the user

  3. The prompt options display a list of all the prompts created, allowing you to drag and drop them to change their order.

  4. Click Done

Multi Response

To allow users to generate multiple responses,

  1. Enable the Allow users to generate multiple responses toggle button in the output settings.

  2. Click Continue to move to the next step.

Step 3: Preview

In the Preview section, a list of sample queries are displayed. If needed, you can add additional queries by clicking the + Add Query button. Once you're ready, click Continue to proceed to the next step.

Step 4: Publish Prompt Agent

In the Publish section, you can finalize and deploy your Prompt agent. Follow the steps outlined below to publish the Prompt agent

  1. Provide the following details

    • Published Version: Select the version of the agent you are publishing.

    • Publish to: Choose who will have access to the agent:

      • Admins: Restrict the agent to Admin users only.

      • Selected User Groups/Users: Specify individual users or groups.

      • Everyone in the Account: Make the agent available to all users.

    • Enablement Type: Define how end users can enable or disable the agent from agent store:

      • Always Enabled: Users cannot disable the agent; it is always active.

      • User's Choice: Users can choose whether to enable or disable the agent as needed.

  2. Click Publish. Once published, your agent is displayed in the Agent list of the Admin console page.

Modify a Prompt Agent

To modify the Prompt agent, follow these steps:

  1. Navigate to the Agents list page and locate the agent you wish to modify.

  2. Click the three dots icon next to the agent’s name. A menu with the following options will appear:

    • Edit – Open and modify the agent's details.

    • Publish/Unpublish – Change the agent's status.

    • Export Agent : The agent's data is packaged into a ZIP file for download, migration, or import into another system.

    • Delete – Permanently remove the agent.

  3. Click on the required option and continue to complete the modifications as needed.

Agent Option

The agent options allow you to export, delete, and un-publish the agent.

User Interaction

Users initiate their interaction by submitting simple queries, such as requesting topic summaries. The agent processes these inputs and provides concise, contextually relevant responses.

Users can also leverage the file upload and URL features, which support multiple formats such as .pdf, .docx, .csv, and .xls. Once a file is uploaded, the system processes its contents and generates the desired output, such as summaries, reports, or insights, based on the data within the file.

When multi-response is enabled, users can use this feature by selecting the Additional Response option. Each prompt is preconfigured, but can be customized to meet specific requirements using a dropdown menu. Users have the flexibility to define the input source for each response, either by using the initial input or by selecting the output from a previous response as input for subsequent responses. This configuration allows for a continuous flow of information, making it ideal for scenarios requiring multiple perspectives or formats.