Skip to content

Lab 2: Invoking WxCC APIs via Webex Connect & AI Agent

Lab Purpose

In Lab 1, you learned two ways to interact with WxCC APIs:

  • Bruno — for direct, developer-style API calls.
  • WxCC Flow Designer — using the HTTP Request node inline within the BasicQueueFlow.

In Lab 2, you will explore a third pattern: invoking WxCC APIs from Webex Connect, triggered by a Webex AI Agent. This pattern lets the AI Agent collect customer details and call the WxCC Callback API during a live customer interaction.

The use case: A customer calls in via PSTN and reaches an AI Agent instead of a live agent. When the customer asks for a human, the AI Agent offers to schedule a callback instead of transferring the customer directly to a queue. If the customer accepts, the AI Agent collects the required details and schedules a callback through the WxCC Callback API.

Lab Objectives

The purpose of this lab is to demonstrate how an Autonomous AI Agent can orchestrate real WxCC API operations through Webex Connect fulfillment flows.

Key objectives include:

  • App Integration: Create a Webex App Integration to authorize Webex Connect against WxCC APIs.
  • Custom Node: Build a reusable Custom Node in Webex Connect that wraps the WxCC Callback API.
  • AI Agent Import: Import a prebuilt Autonomous AI Agent with a callback-scheduling persona and slot filling.
  • Fulfillment Flow: Build a Webex Connect Service Flow to fulfill the AI Agent's request and invoke the WxCC API.
  • Voice Integration: Modify the BasicQueueFlow from Lab 1 to route inbound PSTN calls to the AI Agent.
  • End-to-End Validation: Test the full workflow by calling in via PSTN.
Production vs. Lab Environment

In a real production environment, the AI Agent would typically check the current queue wait time before offering a callback, only suggesting one when the wait is genuinely long. We are skipping this step in our lab because wait times in a lab tenant are always near-zero, which would make the logic uninteresting to test. The AI Agent in this lab will offer a callback unconditionally when a human is requested.

Lab Overview

In this lab you will perform the following tasks:

  1. Create a Webex App Integration for Webex Connect.
  2. Build a reusable WxCC Custom Node in Webex Connect.
  3. Import the "Callback Assistant" Autonomous AI Agent.
  4. Build the Webex Connect fulfillment flow.
  5. Test the AI Agent in Preview.
  6. Modify the BasicQueueFlow to route calls to the AI Agent.
  7. Test the complete scenario end-to-end via PSTN.

Lab 2.1 - Create a Webex App Integration for Connect

In this first step, you will create the OAuth identity that allows Webex Connect to call WxCC APIs on behalf of your organization. Each external client (Bruno, Connect, custom apps) should have its own Integration so that scopes, tokens, and audit trails stay cleanly separated.

Instructions
  1. Navigate to the Webex for Developers website.
  2. Click Log In in the top-right corner and use the admin credentials provided for your POD.
  3. Once logged in, click your avatar in the top-right corner and select My Webex Apps.
  4. Click Create a New App.
  5. Click Create an Integration.
  6. Fill out the integration form with the following values:
    • Integration Name: Connect-WxCC Integration
    • Icon: Pick any icon.
    • App Hub Description: Allows Webex Connect to invoke WxCC Callback APIs.
    • Redirect URI(s): https://ltrcct2009pod{XX}.us.webexconnect.io/callback
    • Scopes: Check the following four scopes: 
      cjp:config
cjp:config_read
cjp:config_write
cjp:user
  7. Click Add Integration at the bottom of the page.
  8. On the confirmation page, copy and save the Client ID and Client Secret somewhere safe. You will need them in Lab 2.2.
Warning

The Client Secret is shown only once. If you lose it, you'll need to regenerate it.

Why a separate Integration?

In Lab 1, you used a Webex Integration for Bruno. Each external client (Bruno, Connect, custom apps) should have its own Integration so that scopes, tokens, and audit trails stay cleanly separated.

Lab 2.2 - Build the WxCC Custom Node in Webex Connect

Webex Connect uses Custom Nodes to wrap REST APIs into reusable flow components. In this section, you will build one that wraps the WxCC scheduled callback API so it can be used in the fulfillment flow.

Create the Webex Connect Service
  1. Go to Control Hub → Contact Center → Overview and under Digital Channels launch [Webex Connect].

  2. Create a new Service for this lab:

    • Click on [Create New Service]
    • Give it a name WxCC API and click on [Create]

  3. From the menu on the left hand side, navigate to Assets > Integrations.

  4. Click Add Integration and select Custom Node.
Download the WxCC Custom Node Icon

Download the custom node icon before configuring the node: wxcc_custom.svg

If the SVG opens in the browser instead of downloading, right-click the link and save the file as wxcc_custom.svg.

  1. Configure the new node:

    Field Value
    Node Icon Upload the downloaded wxcc_custom.svg file
    Node Name WxCC_API
    Description Schedules a callback in Webex Contact Center.
    Node Category WxCC
    Creation Type Create a blank integration

  2. Click [OK]. The new node opens so you can add the API method.

HTTP Connector configuration

HTTP Connector Setup

Confirm the Queue-1 ID

The ScheduleCallback method uses the lab queue Queue-1. Each POD has its own Queue ID, so you must copy the value from your tenant before configuring the Custom Node method body.

  1. In Control Hub, go to Contact Center > Queues.
  2. Open Queue-1.
  3. Copy the ID shown under the queue name.
  4. Save this value temporarily. You will paste it into the queueId field in the next section.

Warning

Do not use another attendee's Queue ID. The value is unique per POD.

Add the ScheduleCallback method
  1. Inside the WxCC_API node, click Add Method

  2. Configure the Request Details section:

    Field Value
    Request Name ScheduleCallback
    Request Timeout (ms) 5000
    Connection Timeout (ms) 5000
    Type Post
    Resource URL https://api.wxcc-us1.cisco.com/v1/callbacks/organization/$(orgId)/scheduled-callback

  3. Click Parse Variables (next to the Resource URL field). Add the following parameter:

    Parameter Parameter Value Type Parameter Value
    orgId Static (paste your Organisation ID from your POD credentials)

  4. Configure the Authorization section:

    Field Value
    Type OAuth 2.0
    Grant Type Authorization Code
    Consumer ID (Client ID from the Webex Integration you created in Lab 2.1)
    Consumer Secret (Client Secret from the Webex Integration you created in Lab 2.1)
    Authorization URL https://webexapis.com/v1/authorize
    Scope cjp:config cjp:config_read cjp:config_write cjp:user
    Access Token URL https://webexapis.com/v1/access_token
    Validity 1209599
    Refresh Token URL https://webexapis.com/v1/access_token
    Client Authentication Send client credentials in body

  5. Under Advance Settings, confirm the following are set:

    Field Value
    Access Token URL Method POST
    Access Token URL Parameter type Body
    Access Token URL Parameters — Name grant_type
    Access Token URL Parameters — Value authorization_code

  6. Click Get Access Token. A Webex login window will appear — sign in using your POD admin credentials. Accept the requested permissions. Your Access Token and Refresh Token will populate automatically.

  7. Configure the Headers section:

    Parameter Parameter Value Type Parameter Value
    Content-Type Static application/json

  8. Configure the Body section. Set the content type to JSON (application/json) and paste in the following:

    {
  "customerName": "$(name)",
  "callbackNumber": "$(callback_number)",
  "timezone": "America/New_York",
  "scheduleDate": "$(date)",
  "startTime": "$(start_time)",
  "endTime": "$(end_time)",
  "queueId": "REPLACE_WITH_QUEUE_1_ID",
  "callbackReason": "$(reason)"
}

  9. Click Parse (inside the Body section). Configure the six dynamic input parameters:

    Parameter Parameter Value Type Field Name
    name Dynamic Name
    callback_number Dynamic Callback Number
    date Dynamic Date
    start_time Dynamic Start Time
    end_time Dynamic End Time
    reason Dynamic Reason

  10. Configure the Response section. Under Configure Node Events, confirm the following is set:

    Node Event Body Response Path Condition Value Node Edge
    Success HTTP Status (blank) starts with 2 Success
    Failure HTTP Status (blank) starts with 4 Error
    Failure HTTP Status (blank) starts with 5 Error

  11. Under Set data to be returned in a flow session, add the following response parameter:

    Parameter Name Body Response Path
    response_body Body $

  12. Click [Save].

Scheduled Callback Method

Custom Node
Scheduled Callback method configuration.

Test the Custom Node
  1. Inside the WxCC_API node, click [Test].
  2. Select the ScheduleCallback method.

  3. Enter sample values:

    Field Example
    Name Jane Smith
    Callback Number 5554443322
    Date Today's date in YYYY-MM-DD format
    Start Time A time at least 30 minutes in the future, HH:mm:ss
    End Time A time at least 30 minutes after start time, HH:mm:ss
    Reason Customer requested a scheduled callback from the AI Agent.

  4. Click [Test] and confirm a successful 2xx response.

Checkpoint

You now have a reusable Webex Connect Custom Node that can schedule callbacks in WxCC. Lab 2.4 will call this node from the AI Agent fulfillment flow.

Troubleshooting
  • 401 / 403: Re-authorize the Custom Node and confirm the Lab 2.1 Integration includes cjp:config, cjp:config_read, cjp:config_write, and cjp:user.
  • 400 - Invalid request: Confirm date, start_time, and end_time use the expected formats and that the start time is at least 30 minutes in the future.
  • Invalid queue: Confirm the queueId matches Queue-1 in your POD.

Lab 2.3 - Import the "Callback Assistant" AI Agent

In this section, you will import the prebuilt Callback Assistant AI Agent. The imported agent already contains the persona, instructions, and the schedule_callback fulfillment action used later in this lab.

Download the Callback Assistant Import File

Download the AI Agent import file: Callback Assistant.json

If the JSON opens in the browser instead of downloading, right-click the link and save the file as Callback Assistant.json.

Import the AI Agent
  1. From Control Hub, navigate to Contact Center and under Quick Links click Webex AI Agent. The Webex AI Agent Studio opens in a new window.
  2. Click Import agent.
  3. Upload the downloaded Callback Assistant.json file.
  4. Complete the import workflow and open the imported Callback Assistant agent.
  5. Confirm the agent is not published yet. You will publish it after the fulfillment flow is connected and tested.
Review the Imported Agent
  1. Open the imported Callback Assistant agent.

  2. Confirm the Welcome message is configured:

    Thanks for calling Webex Contact Center! How can I help you today?

  3. Open the Actions tab and confirm the schedule_callback action exists.

  4. Confirm the action has the following input entities:

    Entity Name Required Purpose
    name Yes Customer name for the scheduled callback
    mobile_number Yes Callback phone number collected by the AI Agent
    date Yes Callback date requested by the caller
    start_time Yes Callback window start time requested by the caller
    end_time Yes Callback window end time requested by the caller
    reason No Optional callback reason

Warning

Do not publish the AI Agent yet. The imported action still needs to be connected to the Webex Connect fulfillment flow you will build in Lab 2.4.

Import AI Agent

AI Agent
Import AI Agent-Callback Assistant

Lab 2.4 - Build the Webex Connect Fulfillment Flow

When the AI Agent triggers schedule_callback, Webex Connect needs a flow to receive the collected callback details, normalize the callback phone number, and invoke the WxCC API via the Custom Node we built in Lab 2.2. In this section you will build that flow.

Action fulfillment flow in Webex Connect
  1. Go to your WxCC API Service in Webex Connect.

  2. Click on [Flows] to create your fulfillment flow.

    • Click on [Create Flow]
    • Give your flow a name schedule_callback
    • Select New Flow under Method
    • Select Start from Scratch
    • Click on [Create]

  3. Configure the Start Node to trigger the flow from the AI Agent:

    • Select the AI Agent Event.

    • Set the SAMPLE JSON to the below:

      {
    "start_time": "",
    "end_time": "",
    "date": "",
    "mobile_number": "",
    "name": "",
    "reason": ""
}
      Info

      The variable names in the JSON sample must match the corresponding input entities in the imported AI Agent action: name, mobile_number, date, start_time, end_time, and reason.

      Note

      This sample JSON is used to parse variables in Flow Builder. The values can stay empty because the AI Agent supplies the real values at runtime.

    • Click on [Parse]

    • Click on [Save]
    Info

    Flows do not autosave, so make sure you save your flow whenever you make edits.

  4. Add an Evaluate node to normalize the phone number before sending it to WxCC.

    The AI Agent may collect the phone number with +, spaces, parentheses, or hyphens. The WxCC API expects the callback number without formatting characters, so this node creates a cleaned variable named phone_Number.

    • Drag and drop an Evaluate node onto the canvas next to the Start Node.
    • Connect the green outlet of the Start Node to the Evaluate node.
    • Double-click the Evaluate node.

    • Paste the following JavaScript into the code editor:

      var phone_Number = "$(n2.aiAgent.mobile_number)".replace(/\+/g, "")
.replace(/ /g, "")
.replace(/\(/g, "")
.replace(/\)/g, "")
.replace(/\-/g, "");

1;

    • Set the Script Output value to 1.

    • Set the Branch Name value to Success.
    • Click [Save].
    • Save the flow.

    Warning

    The n2 prefix refers to the Start Node ID and may be different in your flow. Pick mobile_number from the Input Variables panel if your node ID is different.

  5. Now we'll invoke the WxCC Callback API using the Custom Node we built in Lab 2.2.

    • Drag and drop the WxCC_API node onto the canvas next to the Evaluate node.
    • Connect the green outlet of the Evaluate node to the WxCC_API node.
    • Double-click on the node.
    • Set Method Name to ScheduleCallback.

    • Map the inputs as follows:

      Custom Node Input Mapped Value
      name $(n2.aiAgent.name)
      callback_number $(n5.phone_Number)
      date $(n2.aiAgent.date)
      start_time $(n2.aiAgent.start_time)
      end_time $(n2.aiAgent.end_time)
      reason $(n2.aiAgent.reason)
      Warning

      The n2 and n5 prefixes refer to node IDs and may be different in your flow. Pick the variables from the Input Variables right panel rather than typing them manually. The callback number should come from the phone_Number variable created by the Evaluate node.

    • Click on [Save].

    • Save the flow.

  6. To complete the flow, click on the green outlet at the right of the Custom Node and drag it to the canvas. Release it to open the End dialogue:

    • Set the Node Event value to onSuccess.
    • Set the Flow Result value to 101 - Successfully completed flow [Success].
    • Click on [Save].
    • Open the Transition Actions (Optional) tab for the End node.

    • Click + Add Action and configure the action:

      Field Value
      Time On-enter
      Action Set variable
      Variable result
      Value Callback Scheduled

    • Save the flow.

  7. To finish your fulfillment flow, you need to pass the output variables back to the AI Agent. This is done through the Flow Outcomes.

    Info

    The AI Agent is notified of flow completion. By default, the notification for AI Agent is enabled under the flow setting with the default payload. On flow completion, you can update the payload shared with the AI Agent using the Flow Outcomes.

    • Click on the button at the top right of the flow editor.
    • Click on the Flow Outcomes tab.
    • Open the Last Execution Status Outcome.
    • Confirm Notify AI Agent is selected.
    • Select Enter key and value.

    • Click on + Add New for each variable below. Pick the values from the Input Variables right panel under the appropriate node section:

      Key Value
      transactionID $(transid)
      flowname $(flowname)
      serviceName $(serviceName)
      statuscode 1000
      result $(result)
      response $(n3.response_body)

      Note the numbering for your variables might be different. Make sure response corresponds to the response body from your WxCC_API node.

    • Click on [Save].

    • Save the flow.

Your schedule_callback flow is now completed.

Before leaving the flow editor, make sure you [Make Live] the flow, otherwise it will not be visible to the AI Agent Studio.

Link the flow back to the AI Agent action
  1. Return to AI Agent Studio → open your Callback Assistant agent → Actions tab → open schedule_callback.
  2. Under the Fulfillment section:
    • Use Webex Connect Flow Builder should already be selected.
    • Service: WxCC API
    • Flow: select the freshly published schedule_callback flow.
  3. Click [Save]
Link Flow to Action

AI Agent Action Fulfillment

Lab 2.5 - Test the AI Agent

Before going live with PSTN, validate that the Callback Assistant follows the programmed logic by using the Preview tool in the AI Agent Studio.

  1. In the AI Agent Studio, click the [Preview] button at the top-right of the configuration page.
  2. You can do a preview chat or a preview voice conversation. Initiate a session.

  3. Verification checklist — confirm the Agent successfully executes the following logic:

    • Greeting: Does it warmly greet you and ask how it can help?
    • Intent Detection: When you say "I'd like to talk to a real person", does it offer a callback?
    • Slot Filling: Does it collect your name, phone number, callback date, and callback time window naturally?
    • Confirmation: Does it read the details back before scheduling?
    • Fulfillment: Does it execute schedule_callback and read back the confirmed date and start time?
Troubleshooting your AI Agent

The Sessions option in the left navigation panel of your AI Agent configuration window provides a comprehensive record of all interactions between the AI agent and users. To view session details, click on an individual row in the sessions table. If the session is locked, click the [Decrypt content] button to view the data.

Warning

The [Decrypt content] button appears only if you have decrypt access within the AI Agent Studio. If you do not have access, ask your administrator to grant you decryption privileges.

To debug the Webex Connect flow, use the Flow Debugging tool. Before debugging:

  • Click the at the top-right of the flow editor next to Save.
  • On the Flow Settings screen enable Descriptive Logs in the General tab.
  • Then in the flow designer, click the Bug icon on the far-right navigation pane to open the debug panel.
  • Click Decrypt Logs above the search bar, then select an execution to inspect the per-node data flow.
PUBLISH YOUR AI AGENT

Once you have validated your AI Agent in Preview, you can Publish it.

  • In the AI Agent configuration page, click [Publish] at the top-right corner.
  • Provide a publishing comment in the next dialogue and click Publish.

Your Agent is now operational and ready to be wired into the voice flow.

Lab 2.6 - Connect the AI Agent to the Inbound Call

In this configuration step, you will modify the BasicQueueFlow from Lab 1 so inbound PSTN calls hit the AI Agent first. If the AI Agent successfully schedules a callback, the call ends. If the AI Agent escalates, the call falls back to the original queue path.

Delivering the inbound call to the Callback Assistant
  1. Go to Control Hub → Contact Center → Flows.
  2. Open your flow BasicQueueFlow. Enable the Edit slider at the top of the canvas.

  3. Locate the starting call path:

    NewPhoneContact → PlayMessage1 → QueueContact1 → EndFlow1

  4. From the Activities Library drag and drop a Virtual Agent V2 to the right of the NewPhoneContact event.

    Note the event may appear as NewContact in your flow.

  5. Insert the Virtual Agent V2 node between NewPhoneContact and the existing PlayMessage1 node:

    • Remove the connector from NewPhoneContact to PlayMessage1.
    • Connect NewPhoneContact to Virtual Agent V2.
    • Keep the existing PlayMessage1 → QueueContact1 → EndFlow1 path in place.

  6. Click on the Virtual Agent V2 node and configure the settings below in the Activity Settings panel on the right side of the canvas:

    • Click the pencil on the Activity Label and rename it to AIAgent_Greeter. Click the tick to save.
    • Provide an Activity Description.
    • Under Conversational Experience, set Static Contact Center AI Config.
    • Then in the Contact Center AI Config dropdown, select Webex AI Agent (Autonomous).
    • In the Virtual Agent drop-down menu, select your imported AI Agent (Callback Assistant).
    • In the Decryption Settings, set the Enable decryption slider to ease troubleshooting during debugging. Make sure the Enable decryption slider is also set in the Global Flow Properties panel.

  7. Wire up the output branches:

    • Handled branch → connect to Disconnect Contact (the AI Agent successfully scheduled the callback, so the call can end).
    • Escalated branch → connect to PlayMessage1 so the customer can still reach the original queue path.
    • Leave PlayMessage1 connected to QueueContact1.
    • Leave QueueContact1 connected to EndFlow1.

    Your final flow should be:

    NewPhoneContact → Virtual Agent V2
Virtual Agent V2 / Handled → Disconnect Contact
Virtual Agent V2 / Escalated → PlayMessage1 → QueueContact1 → EndFlow1

  8. Enable the Validation check with the slider in the bottom right of the editor. If there are no errors, click [Publish Flow].

Flow with AI Agent

Final Flow Lab 2

No need to update the Routing Strategy

Because we modified BasicQueueFlow in place, your existing Routing Strategy from Lab 1 already points to it — no further changes are needed.

Lab 2.7 - Test the complete scenario

You can now test the complete scenario end-to-end via PSTN.

  1. Dial your POD's PSTN number (provided by the instructor).

  2. Walk through this dialogue:

    You (Caller) Expected AI Agent Response
    (silence — agent greets first) "Thanks for calling Webex Contact Center! How can I help you today?"
    "I'd like to talk to a real person" "All our agents are currently helping other customers, but I'd be happy to schedule a callback for you. Can I get your full name?"
    "Jane Smith" "Thanks Jane. What's the best phone number to call you back on?"
    "919-555-1234" "What date would you like the callback?"
    "Tomorrow" "What time window works best for you?"
    "Between 3:30 PM and 4:30 PM Eastern" "Just to confirm — Jane Smith at 919-555-1234, tomorrow between 3:30 PM and 4:30 PM. Should I go ahead and schedule the callback?"
    "Yes" (triggers fulfillment) "You're all set, Jane! We'll call you back tomorrow at 3:30 PM. Have a great day!"

  3. Verify the callback was created in WxCC from Bruno.

    Open Bruno and run the scheduled callbacks GET request:

    GET /v1/callbacks/organization/{orgId}/scheduled-callback

    Confirm your newly scheduled callback appears in the response payload.

Warning

As the room environment may be noisy, please avoid using speaker mode for your calls to avoid unintended responses.

Full-Circle Moment

The same GET API you tested in Bruno during Lab 1 now returns a callback that was created end-to-end by an AI Agent talking to you over PSTN and orchestrated by Webex Connect. That's the power of API-driven contact center architecture.

Lab Completion ✅

You've now seen three different ways to invoke WxCC APIs:

Method Lab Best For
Bruno Lab 1 Developer testing, ad-hoc calls
WxCC Flow Designer (HTTP Request) Lab 1 Inline API calls within voice/digital flows
Webex Connect + AI Agent Lab 2 Conversational AI automation with reusable Custom Nodes

At this point, you have built a fully automated Callback Scheduling experience that:

  • [x] Uses a dedicated Webex Integration to securely authorize Webex Connect against WxCC APIs.
  • [x] Wraps the WxCC Callback API in a reusable Custom Node.
  • [x] Greets inbound PSTN callers with an Autonomous AI Agent.
  • [x] Collects customer details via slot filling and triggers a Webex Connect fulfillment flow.
  • [x] Schedules a real callback in WxCC and confirms it back to the caller.
  • [x] Falls back gracefully to the live queue when the AI cannot handle the call.

Congratulations! You have successfully completed Lab 2 and are now ready to move on to the next Lab.

Next Lab: Lab 3 - Event Subscriptions →

Updated at 2026-06-16 22:14:26