Build an AI Agent¶

In the previous tutorial, you learned how to build a RAG application. In this tutorial, you’ll learn how to create an AI agent.

What you'll build¶

PineValley Bank is introducing an AI-powered agent designed to assist users in making informed investment decisions and creating investment accounts tailored to their specific needs. This AI agent utilizes a knowledge base and a Large Language Model (LLM) to analyze user queries, provide personalized investment recommendations, and guide users through the account creation process. By integrating this AI agent into the CustomerService API deployed in WSO2 Micro Integrator, PineValley Bank aims to deliver seamless and efficient customer service, enhancing user satisfaction and experience.

Tools Used by the AI Agent¶

GetBankDocumentsTool: Accesses PineValley Bank's documents from the knowledge base.
CustomerInfoTool: Retrieves customer information from PineValley Bank's internal database.
InvestmentCreationTool: Facilitates the creation of new investment accounts.

Now, it's time to design the AI agent flow. Follow the steps below to create the agent integration.

What you'll learn¶

How to build an AI agent using WSO2 Micro Integrator.

Prerequisites¶

You need Visual Studio Code (VS Code) with the Micro Integrator for VS Code extension installed.

Info

See the Install Micro Integrator for VS Code documentation to learn how to install Micro Integrator for VS Code.
You must have completed the Build a Knowledge Base tutorial under Build your first AI Integration before proceeding. Start the Build a Knowledge Base tutorial if you haven’t completed it yet.

Follow the instructions below to modify the existing project to add the RAG functionality.

Step 1 - Create a new API¶

To develop the above scenario, let's get started with creating a new API in the existing BankIntegration project.

Click on the add new API (+) icon in the APIs in the Micro Integrator Project Explorer to create a new API.
Create a new API named CustomerServiceAPI by giving the name in the Create API form. Click Create to create the API.
In the Service Designer pane, modify the resource Method to POST.

Step 2 - Design the integration¶

Open the Resource View of the newly created API resource by clicking the POST /chat resource under Available resources in the Service Designer.
After opening the Resource View, click on the Start node on the canvas to set an input payload for the integration flow.

Note

Setting an input payload for the integration flow is not mandatory. However, it is recommended, as it will be used to enable expression suggestions, which you will explore in later steps of this tutorial.

Click Add Request, provide the following JSON payload, then click Add. Finally, click Save to complete the input payload setup.
```
{
    "userID": "C567",
    "query": "What can I invest in?"
}
```
Click on the + icon on the canvas to open the Mediator Palette.
Select Agent operation from the AI category in the Mediator Palette.

Fill the form with the following details and click Add to add the RAG Chat operation to the integration flow.

We are going to reuse the same connections that we created in the previous tutorial.

LLM Connection: OPENAI_CONN
Memory Connection: FILE_MEMORY_CONN
User ID: payload.userID (with EX enabled)
Role: PineValley Bank Customer Assistant
Instructions:

Your task is to recommend suitable investment products based on customer data,  
assist with onboarding, and create investment accounts for requested products.  

After creating the investment account, it will be sent to the manager for approval.  
Once it is approved, it will get activated.  

Use tools to retrieve official PineValley Bank documents like brochures and guidelines.  
Provide clear, compliant explanations and step-by-step guidance.  

Always base recommendations on retrieved information and ask for clarification if needed.  
Products include Tax Free Saving Account, High-Interest Savings Account, and more.  

Ensure all interactions are professional, accurate, and regulatory-compliant.

User Query/Prompt: ${payload} (Tip: Click the fx icon next to the User Query/Prompt field, select Payload from the expression panel, choose payload, and click Add to insert this expression.)
Overwrite Message Body: true

Now that we added the Agent operation, we need to add tools to the agent.

To proceed with this tutorial, we will mock the Bank's API to retrieve customer information and create investment accounts. Let's start by creating a mock API to fetch customer details. This mock API will serve as a tool for the AI agent to access customer information during the integration process.
1. Click on the + icon located just after the APIs to add a new API.
2. Create a new API named BankMockAPI by giving the name in the Create API form. Click Create to create the API.
3. In the Service Designer pane of the BankMockAPI, modify the Resource Path to /customerInfo and Method to POST
4. Open the diagram view by clicking on POST /customerInfo under Available resources in the Service Designer.
5. Click on the + icon on the canvas to open the Mediator Palette.
6. Select Payload mediator from the Favourites category in the Mediator Palette.
  Give the following as the payload.
```
{
"userID": "C567",
"name": "John Doe",
"age": 30,
"investmentGoal": "Long-term growth"
}
```
  Click Add to add the payload to the integration flow.
  Now, we need to add a Respond mediator to respond to the client.
  
  Now that we have created the mock API to retrieve customer information, let's proceed to create another mock API for handling investment account creation.
7. Go to Service Designer of the BankMockAPI.
8. In the Service Designer pane, select the + Resource icon to add a new resource.
  Set the following details in the Create Resource form.
  Resource Path: /createInvestmentAccount
  Method: POST.
9. Open the diagram view by clicking on POST /createInvestmentAccount under Available resources in the Service Designer.
10. Click on the + icon on the canvas to open the Mediator Palette.
11. Select Payload mediator from the Favourites category in the Mediator Palette.
  Give the following as the payload.
```
{
"accountID": "INV12345",
"status": "Pending Approval"
}
```
12. Click Add to add the payload to the integration flow.
13. Now, we need to add a Respond mediator to respond to the client.
Now that we have created the mock API to create investment accounts, let's proceed to add the tools to the agent.
Open the POST / resource of the CustomerServiceAPI again.
Click on the + icon located inside the Agent operation to open the Mediator Palette to add a Tool.
Select POST from the HTTP category in the Mediator Palette.
Let's configure the tool with the following details.

Tool Name: CustomerInfoTool
Tool Description: Get customer information from PineValley Bank's internal database
Once you have filled the above details, we need to create a connection to the MockCustomerInfoAPI we created earlier.
Select HTTP from the Add New Connection page.
Fill the form with the following details and click Add to add the connection.

Connection Name: BankMockAPI_CONN
Base URL: http://localhost:8290/bankmockapi

Click Add to add the connection.
In the HTTP POST Tool configuration, set the Relative Path as /customerInfo.

Now, we need to configure the request payload in the tool configuration.
Click on the magic () in the Request Body field.

Note

Clicking on the () action designates the field as an argument for the tool, enabling the AI to populate the field with the necessary values dynamically.
After clicking the magic button, a new input field will appear where you can provide a description for the argument. This description helps the AI understand what needs to be filled for the argument.

Enter the following description in the new field that appears:
```
Provide the customer ID in the following format: {userID: "C567"}
```
Click Add to add the tool.

Next, let's add another tool to get the bank documents from the knowledge base.
Click on the + icon located inside the Agent operation to open the Mediator Palette to add a Tool.
Select Get From Knowledge from the AI category in the Mediator Palette.
Let's configure the tool with the following details.

Tool Name: GetBankDocumentsTool
Tool Description: Get the PineValley bank documents from the knowledge base
Vector Store Connection: KB_CONN
Embedding Model Connection: OPENAI_CONN
Click on the magic () in the Input field.
We can leave the default description as it is.

Next, let's add the final tool to create the investment account.
Click on the + icon located inside the Agent operation to open the Mediator Palette to add a Tool.
Select POST from the HTTP category in the Mediator Palette.
Let's configure the tool with the following details.

Tool Name: InvestmentCreationTool
Tool Description: Create investment account for the customer
Connection: BankMockAPI_CONN
Relative Path: /createInvestmentAccount
Request Body:
Click on the magic () in the Request Body field.
Enter the following description in the new field that appears:
```
Provide the investment account details in the following format: {"userID":"C567", "investmentType":"HISA", "initialDeposit":5000, "investmentGoal":"Long-term growth"}
```
Finally, click Add to add the tool.

Next, add the Respond mediator to finalize the integration flow and send a response back to the client.
Click on the + icon located just after the Agent operation to open the Mediator Palette to add a Respond mediator.

You have successfully updated the integration flow to implement the customer assistance agent.
For reference, you can review the configured API.

Customer Service API

Info

You can view the source view by clicking on the Show Source (</>) icon located in the top right corner of the VS Code.

Design ViewSource View

<?xml version="1.0" encoding="UTF-8"?>
<api context="/customerserviceapi" name="CustomerServiceAPI" xmlns="http://ws.apache.org/ns/synapse">
    <resource methods="POST" uri-template="/">
        <inSequence>
            <ai.agent>
                <connections>
                    <llmConfigKey>OPENAI_CONN</llmConfigKey>
                    <memoryConfigKey>FILE_MEMORY_CONN</memoryConfigKey>
                </connections>
                <sessionId>{${payload.userID}}</sessionId>
                <role>PineValley Bank Customer Assistant</role>
                <instructions>Your task is to recommend suitable investment products based on customer data,
                    assist with onboarding, and create investment accounts for requested products.
                    After creating the investment account, it will be sent to the manager for approval.
                    Once it is approved, it will get activated.
                    Use tools to retrieve official PineValley Bank documents like brochures and guidelines.
                    Provide clear, compliant explanations and step-by-step guidance.
                    Always base recommendations on retrieved information and ask for clarification if needed.
                    Products include Tax Free Saving Account, High-Interest Savings Account, and more.
                    Ensure all interactions are professional, accurate, and regulatory-compliant.
</instructions>
                <prompt>${payload.query}</prompt>
                <responseVariable>ai_agent_1</responseVariable>
                <overwriteBody>true</overwriteBody>
                <modelName>gpt-4o</modelName>
                <temperature>0.7</temperature>
                <maxTokens>4069</maxTokens>
                <topP>1</topP>
                <frequencyPenalty>0</frequencyPenalty>
                <maxHistory>10</maxHistory>
                <toolExecutionTimeout>10</toolExecutionTimeout>
                <tools>
                    <tool name="CustomerInfoTool" template="http_post_tool_1" resultExpression="${vars.http_post_759.payload}" description="Get customer information from PineValley Bank's internal database"/>
                    <tool name="GetBankDocumentsTool" template="ai_getFromKnowledge_tool_1" resultExpression="${vars.ai_getFromKnowledge_571.payload}" description="Get the PineValley bank documents from the knowledge base"/>
                    <tool name="InvestmentCreationTool" template="http_post_tool_2" resultExpression="${vars.http_post_809.payload}" description="Create investment account for the customer"/>
                </tools>
            </ai.agent>
            <respond/>
        </inSequence>
        <faultSequence>
        </faultSequence>
    </resource>
</api>

Step 3 - Run the integration¶

Now that you have updated the integration, it's time to deploy the integration to the Micro Integrator server runtime.

Click the Build and Run icon located in the top right corner of VS Code.

Step 4 - Test the integration service¶

Once the Runtime Services interface is open, select the CustomerServiceAPI, and click the Try It button.

Select the / resource and click Try it Out to test the API.

Let's first ask the agent about the investment options available.

{
    "userID": "001",
    "query": "What can I invest in?"
}

Give the above payload in the Request Body and click Execute.

You should see a response similar to the one below.

{
"content": "Based on the PineValley Bank documents, here are some investment options available to you:\n\n1. **High-Interest Savings Account (HISA)**:\n   - Offers a competitive interest rate of 3.5% per annum.\n   - Minimum balance required is $1,000.\n   - Provides unlimited and free withdrawals.\n   - Suitable for US residents aged 18 and older.\n\n2. **Tax-Free Savings Account (TFSA)**:\n   - Allows tax-free growth on savings and investment earnings.\n   - Contributions are not tax-deductible, but all withdrawals are tax-free.\n   - The contribution limit for 2025 is $7,000 (subject to change annually).\n   - Suitable for Canadian residents aged 18 and older, with a valid SIN.\n   - Investment options include GICs, stocks, ETFs, mutual funds, and cash deposits.\n\nIf you have specific preferences or goals, please let me know so I can provide more tailored recommendations.",
"tokenUsage": {
    "inputTokenCount": 21688,
    "outputTokenCount": 221,
    "totalTokenCount": 21909
},
"finishReason": "STOP",
"toolExecutions": [
    {
    "request": {
        "id": "call_U209o5wZyxrHTas7oNWn8vdZ",
        "name": "ai_getFromKnowledge_tool_1",
        "arguments": "{\"input\":\"investment options PineValley Bank\"}"
    },
    "result": "[{\"score\":0.7656985474367457,\"embeddingId\":\"dd2d5a41-baed-4755-be81-82d5f49599a9\",\"embedding\":[...],\"embedded\":{\"text\":\"Article: Overview of Tax-Free Savings Account (TFSA)\\nDocument Title: PineValley Bank – TFSA Product Brochure\\nLast Updated: March 2025\\nDocument Type: Product Brochure\\nApplies To: All eligible Canadian residents aged 18 and older\\n\\nProduct Summary\\nA Tax-Free Savings Account (TFSA) allows you to grow your savings and investment earnings tax-free. Contributions are not tax-deductible, but withdrawals (including any gains) are completely tax-free.\\n\\nWho Is It For?\\n\\n- Individuals looking to save or invest with flexibility\\n- Customers with short- or long-term financial goals\\n- Those who want to earn interest, dividends, or capital gains tax-free\\n\\nKey Features\\n\\n- Contribution Limit (2025): $7,000 (subject to change annually by CRA)\\n- Unused Contribution Room: Carries forward indefinitely\\n- Withdrawals: Tax-free and can be re-contributed the following year\\n- Investment Options: GICs, stocks, ETFs, mutual funds, cash deposits\\n- Eligibility: Canadian residents, 18+ years old, with valid SIN\\n\\nRegulatory Guidelines\\n\\n- Must comply with Canada Revenue Agency (CRA) limits and rules\\n- Over-contributions are subject to a 1% monthly penalty\\n- TFSA earnings do not affect federal income-tested benefits\\n\\nOnboarding Requirements\\nTo open a TFSA with PineValley Bank:\\n\\n1. Provide proof of age (e.g., driver’s license or passport)\\n2. Upload your Social Insurance Number (SIN)\\n3. Complete a Risk Tolerance Questionnaire\\n4. Agree to TFSA terms & conditions via e-signature\\n\\nEstimated completion time: 15–20 minutes\\n\\nFAQs\\n\\nQ: Can I open more than one TFSA account?\\nA: Yes, but your total contributions across all accounts cannot exceed your CRA limit.\\n\\nQ: What happens if I withdraw money?\\nA: Withdrawals are tax-free and re-added to your contribution room the next year.\\n\\nQ: Can I transfer my TFSA from another institution?\\nA: Yes. We support direct TFSA transfers without affecting your contribution room.\",\"metadata\":{\"index\":\"0\"}}}]"
    }
]
}

In the response above, the agent utilized the GetBankDocumentsTool to retrieve information about the investment options available at PineValley Bank. It then formulated an answer based on the retrieved data to address the user's query effectively.

Now let's ask the agent to create an investment account for High-Interest Savings Account.

{
    "userID": "001",
    "query": "I want to create a High-Interest Savings Account with an initial deposit of $5000."
}

Give the above payload in the Request Body and click Execute.

You should see a response similar to the one below.

{
"content": "Your High-Interest Savings Account (HISA) with an initial deposit of $5,000 has been successfully created and is currently pending approval. Here are the details:\n\n- **Account ID**: INV12345\n- **Status**: Pending Approval\n\nOnce the account is approved by the manager, it will become active. If you have any further questions or need assistance, feel free to ask!",
"tokenUsage": {
    "inputTokenCount": 48122,
    "outputTokenCount": 126,
    "totalTokenCount": 48248
},
"finishReason": "STOP",
"toolExecutions": [
    {
    "request": {
        "id": "call_ZuOsW99DaaQLv8LJd49jj55C",
        "name": "http_post_tool_2",
        "arguments": "{\"requestBodyJson\":\"{\\\"userID\\\":\\\"001\\\",\\\"investmentType\\\":\\\"HISA\\\",\\\"initialDeposit\\\":5000,\\\"investmentGoal\\\":\\\"Long-term growth\\\"}\"}"
    },
    "result": "{\"accountID\":\"INV12345\",\"status\":\"Pending Approval\"}"
    }
]
}

Here is the response from the agent extracted from the above response:

Your High-Interest Savings Account (HISA) with an initial deposit of $5,000 has been successfully created and is currently pending approval. Here are the details:  
- **Account ID**: INV12345  
- **Status**: Pending Approval  

Once the account is approved by the manager, it will become active. If you have any further questions or need assistance, feel free to ask!

In the response above, the agent utilized the InvestmentCreationTool to create a High-Interest Savings Account with an initial deposit of $5000. The account is currently pending approval, and once approved, it will become active. The agent also provided the account ID for reference.
This demonstrates how the agent can assist users in creating investment accounts and provide real-time updates on the status of their requests.

You have now learned how to create an AI agent to enhance your integration flow with automating tasks using AI.

What's Next?¶

You have now completed this tutorial series, where you learned how to build a complete integration flow step by step. Throughout this journey, you have explored how to create AI chatbot, build a knowledge base, build RAG application, and develop an AI agent to enhance your integration flow.

Now, you can explore advanced integration scenarios, applying what you’ve learned to real-world use cases and expanding your expertise in WSO2 Micro Integrator.

Try more tutorials and examples.