Skip to content

Develop Your First AI Agent

Overview

In this guide, you will:

  • Create a simple AI agent that provides math tutoring assistance.
  • Use WSO2 Integrator: BI to develop the AI agent integration.
  • Push the AI agent to Devant from WSO2 Integrator: BI. This will automatically build and deploy it to the development environment.
  • Test the AI agent by sending prompts.
  • Once verified, promote it to the production environment and expose it as an API.

Prerequisites

  1. GitHub account: Create a GitHub account if you don't have one already.
  2. If signing in to Devant for the first time, create an organization to begin with.
  3. VS Code: Install VS Code if you don't have it installed already.

Step 1: Create the new integration

  1. Import a new repository to begin your integration. Refer Import a Repository for more details.
  2. Select the Technology as WSO2 Integrator: BI.
  3. Choose the Integration Type as AI Agent and click Create.

This redirects you to the Create New Integration in VS Code page.

Step 2: Install the WSO2 Integrator: BI extension and open the integration

  1. Install WSO2 Integrator: BI extension. Refer Install and Set Up WSO2 Integrator: BI for more details.
  2. Click Develop in WSO2 Integrator: BI. This will clone your integration project and open it in BI.

Step 3: Develop the AI Agent in WSO2 Integrator: BI

To build your AI Agent using WSO2 Integrator: BI, refer to the AI Agent Integration guide on the Learn page.

Step 4: Push to Devant

  1. Click the Source Control icon on the sidebar.
  2. Click + to stage all changes.
  3. Add an appropriate commit message and commit.
  4. Click Sync Changes to push the changes to remote.
  5. Go back to the Design view using the back arrow in the top left corner.
  6. From the right side panel, click the View in Devant to view this integration in Devant.

Step 5: Test AI Agent

  1. Once you push the changes, the overview page of the Devant AI Agent will automatically refresh and show you the Latest Commit and automatically build and show the Build Status.

    Note

    The build process may take some time. Once complete, the build status changes to Success. You can see the Build History by clicking Build in the left navigation.

  2. Once the Build Status shows Build completed, it will be automatically deployed into the development card.

  3. Then, click the Configure to Continue button, enter the configurable values, and click Apply.

  4. Send the prompt What is 87,657 * 67,997 to test your Math Tutor agent. Even though typically, LLMs are bad at large multiplication, this agent has tool support to give an accurate answer.

    Ballerina Integrator Icon

  5. After successfully testing, you can promote your AI Agent to production by clicking the Promote button.

  6. Once deployed to production, you can access your AI Agent through the API endpoints and embed it in your applications.