# Adding a Webex AI Agent to a Webex Contact Center flow

Systems: Webex Contact Center

For Webex Contact Center administrators adding AI self-service to an existing voice entry point.

Canonical: https://warmtransfer.net/guides/wxcc-virtual-agent-setup

Last verified: 2026-09-30

Webex AI Agent Studio provides scripted agents that follow defined rules and autonomous agents that pursue goals through knowledge bases and actions[^5]. Placing a Webex AI Agent in front of a Webex Contact Center voice flow allows automated self-service before escalating callers to human queues[^54].

## Before you start

- Confirm the organization holds the AI Agent add-on for the Flex 3.0 Contact Center license[^30].
- Confirm your administrator role in Webex Control Hub: Full Administrators receive full access including transcripts, Contact Center Service Administrators receive full access without transcripts, and Read-only Administrators and Supervisors receive read-only access without transcripts[^42].
- Ensure an inbound telephony entry point and associated queue are already provisioned[^58][^59].
- Check tenant capacity against the organization limit of 100 AI agents, which counts scripted and autonomous agents together[^4].

## What changes by situation

- What kind of AI agent are you building? Scripted agent; Autonomous agent.
- Where do escalated callers go? Single static queue; Queue selected by escalation trigger.
- What context should the human agent see upon escalation? Conversation transcript only; Transcript plus AI-generated summary.

## Step 1: Access AI Agent Studio

### Do
Open AI Agent Studio from Webex Control Hub by navigating to Services > Contact Center > Customer experience > AI Agent[^41]. AI Agent Studio must be opened through Control Hub rather than Webex Connect to access voice preview[^57].

### Verify
Confirm the AI Agent Studio dashboard loads and displays the + Create agent button[^11].

### Rollback
Suggested rollback: Close the AI Agent Studio browser tab.

## Step 2: Create the agent

**Scripted agent**

### Do
Click + Create agent and select either a template or Start from scratch[^11]. Scripted templates include Doctor's Appointment and Track Package[^45]. Use the Webex AI Pro 2.0 engine only if you can provide at least 10 representative utterances per intent[^35]. Cisco states that Webex AI Pro 1.0, Webex AI Pro-US 1.0, and Webex AI Pro-Europe 1.0 will be deprecated in the near future without a published date[^17].

### Verify
Suggested check: Verify the agent appears on the AI Agent Studio dashboard listed as a scripted agent.

### Rollback
Delete the agent from the dashboard agent menu, noting that deleting an agent permanently removes it from the system[^14].

**Autonomous agent**

### Do
Click + Create agent, choose a template or Start from scratch, and enter an agent name, a system ID, and an AI engine[^11]. Autonomous templates include Doctor's Appointment, Track Package, Cumulus Bank, and Cumulus Airline[^45]. The Webex AI Pro-US and Webex AI Pro-Europe engines support English only, whereas global Webex AI Pro engines are multilingual[^34]. Cisco states that Webex AI Pro 1.0, Webex AI Pro-US 1.0, and Webex AI Pro-Europe 1.0 will be deprecated in the near future without a published date[^17].

### Verify
Suggested check: Verify the agent appears on the dashboard listed as an autonomous agent showing the selected AI engine.

### Rollback
Delete the agent from the dashboard agent menu, noting that deleting an agent permanently removes it from the system[^14].

## Step 3: Populate agent knowledge and intents

**Scripted agent**

### Do
Configure intents for each customer request, providing at least 10 representative utterances per intent when using the Webex AI Pro 2.0 engine[^35]. Retain the 3 system intents: Default Fallback, Help, and Talk to an agent[^36]. Ensure the Default Fallback intent contains no slots and retains the default fallback template key[^19].

### Verify
Type queries into the chat preview to confirm each intent triggers its expected response, and verify an out-of-scope query triggers the Default Fallback response[^10][^19].

### Rollback
Suggested rollback: Remove or edit the custom intents and utterances before publishing.

**Autonomous agent**

### Do
Click the Knowledge icon in the left navigation, click Add source, and select Files, Articles, or Websites[^29]. Supported file types are.xlsx,.xls,.csv,.pdf,.docx,.doc, and.txt[^25]. A knowledge base supports up to 100 files and 2 GB total, with individual files up to 10 MB,.txt files up to 2 MB, and PDFs up to 300 pages[^26]. Do not upload PCI data, personally identifiable information, or protected health information[^28]. Navigate to Configurations > Knowledge on the agent and link a knowledge base, choosing one in the same language as the agent[^27].

### Verify
In chat preview, submit a query whose answer exists exclusively in the uploaded knowledge source and verify the agent returns the correct information[^10].

### Rollback
Suggested rollback: Unlink the knowledge base on the Configurations > Knowledge tab or remove the added knowledge sources.

## Step 4: Configure backend fulfillment

**Scripted agent**

### Do
Configure custom events for voice fulfillment, or use HTTP nodes which are supported across both channels[^38]. Identify which response requires fulfillment by parsing session metadata at `$.model_state.template_key`[^39]. Configure the Virtual Agent V2 activity's State Event settings in Flow Designer to pass custom events to the agent[^56].

### Verify
Suggested check: Trigger a fulfillment intent in preview and verify the response reflects data from the backend rather than a static placeholder.

### Rollback
Suggested rollback: Remove the custom event or HTTP fulfillment configuration from the intent response.

**Autonomous agent**

### Do
Add actions to the autonomous agent, configuring up to a maximum of 10 actions[^3]. For each action, select a service and flow configured in Webex Connect, configure slot filling entities with a name, type (string, integer, number, boolean, array, or object), description, and examples[^1][^2]. If building from an autonomous template, create the required fulfillment flows in the linked Webex Connect tenant before deploying[^46].

### Verify
Suggested check: Trigger the action in preview, verify slot-filling prompts prompt for missing entities, and verify the action executes.

### Rollback
Suggested rollback: Disable or delete the configured action in the agent's Actions tab.

## Step 5: Configure escalation behavior

**Scripted agent**

### Do
Retain the Talk to an agent intent, which automatically initiates a transfer to a human agent when invoked[^44]. Add representative phrases for human escalation requests; changing the response template key does not alter the handover outcome[^43]. Keep distinct intents for requests intended to escalate to different queues, as the previous intent can be extracted from metadata during flow routing[^37].

### Verify
In chat preview, type a transfer phrase such as "I want to speak to someone" and confirm the agent triggers human handover[^43].

### Rollback
Suggested rollback: Remove the added handover utterances from the Talk to an agent intent.

**Autonomous agent**

### Do
Ensure the default Agent handover action toggle is enabled[^23]. Add custom transfer actions for distinct escalation queues, defining an action name, a natural-language transfer condition, transfer visibility set to announced or silent, and optional input entities[^13]. In an announced transfer the agent informs the customer about the transfer, while in a silent transfer it transfers without mentioning it[^8]. Custom transfer actions exit the Virtual Agent V2 activity through the escalated path with metadata[^12].

### Verify
Suggested check: Trigger each custom transfer condition in preview and verify the agent initiates escalation.

### Rollback
Suggested rollback: Disable the custom transfer actions or toggle the default Agent handover action off.

## Step 6: Test and publish the agent

### Do
Test the agent in chat preview and voice preview[^10]. Voice preview displays a live transcript and supports barge-in[^10]. Ensure the AI transparency message remains enabled, which is the default setting for new agents and plays before the voice welcome message[^7]. Click Publish on the agent configuration page and assign a version name[^32].

### Verify
Open the History page in AI Agent Studio and confirm the change record for the published version is present[^32].

### Rollback
Suggested rollback: Edit the agent configuration and publish a new version to overwrite changes.

## Step 7: Add Virtual Agent V2 to the voice flow

### Do
In Flow Designer, place a Virtual Agent V2 activity into the main flow[^50]. Do not place the Virtual Agent V2 activity after a Queue Contact activity or in the Event Flow canvas[^55]. Select Webex AI Agent Scripted or Webex AI Agent Autonomous to match your agent type, then select the published agent from the Virtual Agent list[^50]. Either configure language and voice overrides on the activity, or ensure Global_Language and Global_VoiceName variables are defined in the flow[^52].

### Verify
Suggested check: Verify the Virtual Agent V2 activity displays no validation errors and lists the selected agent name.

### Rollback
Suggested rollback: Delete the Virtual Agent V2 activity from the flow canvas and reconnect previous nodes.

## Step 8: Wire Handled and Errored paths

### Do
Connect the Handled output path to a disconnect or post-call survey activity[^54]. Connect the Errored output path to a fallback queue or error handling path[^54]. If differentiated error handling is required, branch using a Case activity on `VirtualAgentV2.ErrorCode`, which provides values including `no_error`, `max_no_input`, and `system_error`[^9][^51].

### Verify
Suggested check: Run flow validation in Flow Designer to ensure no unconnected output pins exist on the Virtual Agent V2 activity.

### Rollback
Suggested rollback: Disconnect the Handled and Errored output paths and rewire to original flow logic.

## Step 9: Route the Escalated path

**Scripted agent + Queue selected by escalation trigger**

### Do
Connect the Escalated output path of the Virtual Agent V2 activity to a Parse activity[^31][^54]. Set the Parse activity input variable to `VirtualAgentV2.MetaData`, content type to JSON, and extract the previous intent into a flow variable[^31][^37][^53]. Connect the Parse activity to a Case activity branching on the extracted intent value, with each branch leading to a Queue Contact activity[^9][^33]. Provide a default Case branch routed to a fallback queue[^9][^33].

### Verify
In Flow Designer debug mode, trace an escalated call execution to confirm `VirtualAgentV2.MetaData` is parsed and routes through the matching Case branch[^20].

### Rollback
Suggested rollback: Remove the Parse and Case activities and wire Escalated directly to a single static Queue Contact activity.

**Autonomous agent + Queue selected by escalation trigger**

### Do
Connect the Escalated output path of the Virtual Agent V2 activity to a Parse activity[^31][^54]. Set the Parse activity input variable to `VirtualAgentV2.MetaData`, content type to JSON, and extract `$.escalation_trigger` into a flow variable[^18][^31][^53]. Connect the Parse activity to a Case activity that branches on the extracted trigger value, routing each branch to a Queue Contact activity[^9][^33]. Include a default Case branch pointing to a general fallback queue[^9][^33].

### Verify
In Flow Designer debug mode, trace an escalated call execution to confirm `VirtualAgentV2.MetaData` is parsed and routes through the matching Case branch[^20].

### Rollback
Suggested rollback: Remove the Parse and Case activities and wire Escalated directly to a single static Queue Contact activity.

**Single static queue**

### Do
Connect the Escalated output path of the Virtual Agent V2 activity directly to a Queue Contact activity configured with a static queue[^33][^54].

### Verify
Suggested check: Confirm in Flow Designer that the Escalated pin links directly to the Queue Contact activity.

### Rollback
Suggested rollback: Disconnect the Escalated output path from the Queue Contact activity.

## Step 10: Configure agent desktop handoff context

**Conversation transcript only**

### Do
In Flow Designer, select the Virtual Agent V2 activity and check Enable Conversation Transcript[^16]. Ensure the human agents' desktop layout includes the IVR Transcript widget, which appears only when an administrator has configured it[^24]. Optionally, mark parsed flow variables as agent-viewable so they display on the desktop interaction pane[^6].

### Verify
Suggested check: Place a test call, escalate to an agent, and verify the IVR Transcript widget on the desktop displays the conversation transcript.

### Rollback
Deselect the Enable Conversation Transcript checkbox on the Virtual Agent V2 activity[^16].

**Transcript plus AI-generated summary**

### Do
Select the Virtual Agent V2 activity and check Enable Conversation Transcript[^16]. Ensure the desktop layout includes the IVR Transcript widget[^24]. Confirm agents have the A-FLEX-AI-ASST add-on license and AI Assistant access enabled in Desktop Experience settings[^48]. In Control Hub, navigate to Services > Contact Center > Desktop Experience > AI Features, open Generated Summaries, and turn on Virtual agent transfer summaries for all queues or selected escalation queues[^47]. If using a custom desktop layout, add the `ai-assistant` component to `advancedHeader`[^49].

### Verify
Suggested check: Answer an escalated test call on the desktop and verify both the IVR Transcript widget and the AI Assistant transfer summary appear.

### Rollback
In Control Hub under Services > Contact Center > Desktop Experience > AI Features > Generated Summaries, turn off Virtual agent transfer summaries or remove the escalation queues[^47].

## Step 11: Validate, publish, and map the flow

### Do
Validate the flow in Flow Designer and correct any errors[^60]. Publish the flow with publish notes and assign a version label such as Test or Live[^22]. In Control Hub under Services > Contact Center > Customer experience > Channels, select your inbound telephony entry point and assign the flow and desired version label[^61][^62][^63][^22].

### Verify
Place a phone call to the entry point's mapped PSTN number and confirm the AI transparency message plays followed by the agent welcome message[^64][^7].

### Rollback
Suggested rollback: Move the Live label back to the previous flow version or point the channel setting to a previous version label.

## Step 12: Perform end-to-end testing

### Do
Place test calls through the entry point covering each outcome: a call resolved entirely by self-service (Handled), each escalation reason (Escalated), and caller silence (Errored with `max_no_input`)[^51][^54]. Use Flow Designer's debug tool to trace the execution paths, variable outputs, and activity behavior for each test call[^20].

### Verify
Open AI Agent Studio Sessions, apply filters for "Agent handover occurred" and "Error occurred", and match the session logs to your test calls[^40]. Hide test sessions from future review views if desired[^40].

### Rollback
Suggested rollback: Disconnect the entry point from the published flow version or adjust the routing flow configuration.

## Applicability

Applies to: Cisco Webex AI Agent Studio, Cisco Webex Contact Center, and Cisco Webex Contact Center Agent Desktop. Deployments: multi-tenant. Sources checked 2026-09-30. Licensing claims apply to Webex Contact Center Flex 3.0[^30][^48]. Autonomous engine language restrictions apply specifically to the Webex AI Pro-US and Webex AI Pro-Europe engines[^34].

## What remains uncertain

The exact JSON schema of `VirtualAgentV2.MetaData` is not covered by the sources below, beyond the presence of `$.escalation_trigger` and `$.model_state.template_key`. Whether virtual agent transfer summaries support scripted agents, autonomous agents, or both is not covered by the sources below. Whether the Virtual Agent V2 placement limitations and language variable constraints apply equally across all engine types is not covered by the sources below. A procedure to roll back an AI agent to an earlier published version in AI Agent Studio is not covered by the sources below. The specific deprecation dates for Webex AI Pro 1.0, Webex AI Pro-US 1.0, and Webex AI Pro-Europe 1.0 are not covered by the sources below.

## Sources

[^1]: An autonomous agent action's fulfillment is configured by choosing a service and flow configured in Webex Connect; MCP client and custom transfer are the other action configuration types. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), Add actions to autonomous AI agent section, fulfillment. Checked 2026-09-30.
[^2]: Action slot filling is configured as entities with a name, a type of string, integer, number, boolean, array or object, a description and examples, entered in a table or a JSON editor. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), Add actions to autonomous AI agent section, slot filling. Checked 2026-09-30.
[^3]: An autonomous agent can have at most 10 actions. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), Add actions to autonomous AI agent section. Checked 2026-09-30.
[^4]: An organisation can create up to 100 AI agents in AI Agent Studio, counting scripted and autonomous agents together. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), Create an AI agent section. Checked 2026-09-30.
[^5]: AI Agent Studio offers autonomous agents, which pursue defined goals using predefined actions and a knowledge base, and scripted agents, which follow predefined rules and answer from a user-created training corpus. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), Overview of AI agent types. Checked 2026-09-30.
[^6]: Flow variables marked agent viewable are displayed on Agent Desktop during the interaction. Source: [Build and manage flows with Flow Designer](https://help.webex.com/en-us/article/nhovcy4/Build-and-manage-flows-with-Flow-Designer), Flow variables, agent viewable setting. Checked 2026-09-30.
[^7]: The AI transparency message is enabled by default for new agents and plays before the welcome message on voice or in the first response on digital. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), AI transparency section. Checked 2026-09-30.
[^8]: In an announced transfer the AI agent tells the customer about the transfer; in a silent transfer it transfers without mentioning it. Source: [Multi-agent orchestration](https://help.webex.com/en-us/article/5a07xcb/Multi-agent-orchestration), Announced and silent transfers section. Checked 2026-09-30.
[^9]: Flow Designer's Case activity branches on one variable against multiple values, and the Condition activity branches true or false on an expression. Source: [Build and manage flows with Flow Designer](https://help.webex.com/en-us/article/nhovcy4/Build-and-manage-flows-with-Flow-Designer), Case and Condition activities. Checked 2026-09-30.
[^10]: An agent can be tested in AI Agent Studio's chat preview by typing a query, and voice preview shows a live transcript and supports barge-in. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), Preview AI agent section. Checked 2026-09-30.
[^11]: An agent is created from the dashboard with + Create agent, either from a template or from scratch, and an autonomous agent requires an agent name, a system ID and an AI engine. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), Create an autonomous AI agent section. Checked 2026-09-30.
[^12]: A custom transfer action exits the Virtual Agent V2 activity through its escalated path with metadata the flow can use to orchestrate the next step. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), Configure custom transfer action section. Checked 2026-09-30.
[^13]: A custom transfer action has an action name, a natural-language transfer condition, a transfer visibility setting for announced or silent transfer, and optional input entities. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), Configure custom transfer action section. Checked 2026-09-30.
[^14]: Deleting an AI agent in AI Agent Studio permanently removes it from the system. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), Dashboard agent menu, Delete. Checked 2026-09-30.
[^15]: Digital channels use the AI Agent node in Webex Connect rather than Virtual Agent V2, and the node's success outcomes include onAgentHandover. Source: [Use AI agents for customer interactions](https://help.webex.com/en-us/article/s0qro1/Use-AI-agents-for-customer-interactions), Configure AI agent for digital channels section. Checked 2026-09-30.
[^16]: Checking Enable Conversation Transcript on the Virtual Agent V2 activity lets Agent Desktop display the transcript of the conversation between the virtual agent and the caller. Source: [Configure Virtual Agent-Voice in Webex Contact Center](https://help.webex.com/en-us/article/n6gaghu/Configure-Virtual-Agent-Voice-in-Webex-Contact-Center), Virtual Agent V2 activity settings, Conversation Transcript. Checked 2026-09-30.
[^17]: Cisco states that Webex AI Pro 1.0, Webex AI Pro-US 1.0 and Webex AI Pro-Europe 1.0 will be deprecated in the near future, without a date. Source: [Understand AI engines for AI agents](https://help.webex.com/en-us/article/ne6s80cb/Understand-AI-engines-for-AI-agents), Autonomous AI agent engines section, deprecation note. Checked 2026-09-30.
[^18]: Cisco's multi-agent orchestration guidance identifies which transfer fired by reading $.escalation_trigger from the Virtual Agent V2 metadata. Source: [Multi-agent orchestration](https://help.webex.com/en-us/article/5a07xcb/Multi-agent-orchestration), Custom transfer to human agent section. Checked 2026-09-30.
[^19]: The Default Fallback intent handles out-of-scope queries, should have no slots, and must use the default fallback template key. Source: [Understand intents, entities, and responses in AI Agent Studio](https://help.webex.com/en-us/article/sz02k8/Understand-intents,-entities,-and-responses-in-AI-Agent-Studio), Default intents section, Default Fallback Intent. Checked 2026-09-30.
[^20]: Flow Designer has a debug capability that traces flow executions and shows activity outputs and variable values. Source: [Build and manage flows with Flow Designer](https://help.webex.com/en-us/article/nhovcy4/Build-and-manage-flows-with-Flow-Designer), Debug flows section. Checked 2026-09-30.
[^21]: Because the channel runs the flow version carrying the selected label, moving that label back to the previous version or pointing the channel at another label should withdraw a newly published AI agent flow (inferred). Source: [Build and manage flows with Flow Designer](https://help.webex.com/en-us/article/nhovcy4/Build-and-manage-flows-with-Flow-Designer), Validate and publish flow; version labels. Checked 2026-09-30.
[^22]: A validated flow is published with publish notes and version labels Latest, Dev, Test or Live, and the channel's routing flow setting chooses which labelled version runs. Source: [Build and manage flows with Flow Designer](https://help.webex.com/en-us/article/nhovcy4/Build-and-manage-flows-with-Flow-Designer), Validate and publish flow; version labels. Checked 2026-09-30.
[^23]: An autonomous agent's Agent handover action is enabled by default, letting it escalate to a human agent, and can be disabled with its toggle. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), Add actions to autonomous AI agent section. Checked 2026-09-30.
[^24]: Agent Desktop's IVR Transcript widget is a view-only display of the conversation between the virtual assistant and the caller, and appears only if an administrator has configured it. Source: [Get started with Agent Desktop](https://help.webex.com/en-us/article/itebdl/Get-started-with-Agent-Desktop), IVR Transcript widget section. Checked 2026-09-30.
[^25]: Knowledge base file uploads accept .xlsx, .xls, .csv, .pdf, .docx, .doc and .txt files. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), Knowledge base, supported file types. Checked 2026-09-30.
[^26]: A knowledge base holds at most 100 files and 2 GB in total, each file at most 10 MB, .txt files at most 2 MB, and PDFs at most 300 pages. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), Knowledge base, file limits. Checked 2026-09-30.
[^27]: A knowledge base is linked to an agent on the Configurations > Knowledge tab, and Cisco recommends choosing one in the same language as the agent. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), Configure knowledge base section. Checked 2026-09-30.
[^28]: Cisco instructs administrators not to upload PCI data, personally identifiable information or protected health information to a knowledge base, and makes the customer responsible for uploaded content. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), Knowledge base, content restrictions note. Checked 2026-09-30.
[^29]: A knowledge source is added from the Knowledge icon in the left navigation with Add source, choosing Files, Articles or Websites. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), Create knowledge source for AI agent section. Checked 2026-09-30.
[^30]: For Webex Contact Center, using Webex AI Agent requires the AI Agent add-on for the Flex 3.0 Contact Center license. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), Prerequisites / licensing section near the top of the article. Checked 2026-09-30.
[^31]: Flow Designer's Parse activity takes an input variable, a content type such as JSON, and JSONPath expressions that map values into flow variables. Source: [Build and manage flows with Flow Designer](https://help.webex.com/en-us/article/nhovcy4/Build-and-manage-flows-with-Flow-Designer), Parse activity. Checked 2026-09-30.
[^32]: An agent is published with Publish on its configuration page and a version name, and the History page shows configuration change records. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), Publish AI agent section; View sessions and history section. Checked 2026-09-30.
[^33]: The Queue Contact activity places a contact in a queue chosen either statically or from a flow variable that yields a valid queue ID. Source: [Build and manage flows with Flow Designer](https://help.webex.com/en-us/article/nhovcy4/Build-and-manage-flows-with-Flow-Designer), Queue Contact activity. Checked 2026-09-30.
[^34]: For autonomous agents, the Webex AI Pro-US and Webex AI Pro-Europe engines are for US and EU customers respectively and support English only, while the global Webex AI Pro engines are multilingual. Source: [Understand AI engines for AI agents](https://help.webex.com/en-us/article/ne6s80cb/Understand-AI-engines-for-AI-agents), Autonomous AI agent engines section. Checked 2026-09-30.
[^35]: A scripted agent using the Webex AI Pro 2.0 engine needs at least 10 representative utterances per intent. Source: [Understand AI engines for AI agents](https://help.webex.com/en-us/article/ne6s80cb/Understand-AI-engines-for-AI-agents), Scripted AI agent engines section, Webex AI Pro 2.0. Checked 2026-09-30.
[^36]: A scripted agent has three system intents: Default Fallback, Help, and Talk to an agent. Source: [Understand intents, entities, and responses in AI Agent Studio](https://help.webex.com/en-us/article/sz02k8/Understand-intents,-entities,-and-responses-in-AI-Agent-Studio), Default intents section. Checked 2026-09-30.
[^37]: For a scripted agent on voice, escalation leaves the Virtual Agent V2 activity through the Escalated output, and the agent's previous intent can be read from the metadata to choose a queue with branch or case logic. Source: [Configure fulfillment for scripted AI agents](https://help.webex.com/en-us/article/mzpuseb/Configure-fulfillment-for-scripted-AI-agents), Handover / escalation section. Checked 2026-09-30.
[^38]: Scripted agent fulfillment uses a Webex Connect flow for digital channels, HTTP nodes for both channels, and custom events for voice channels. Source: [Configure fulfillment for scripted AI agents](https://help.webex.com/en-us/article/mzpuseb/Configure-fulfillment-for-scripted-AI-agents), Fulfillment options section. Checked 2026-09-30.
[^39]: For scripted agent fulfillment, the response name is obtained by parsing the agent session metadata at $.model_state.template_key. Source: [Configure fulfillment for scripted AI agents](https://help.webex.com/en-us/article/mzpuseb/Configure-fulfillment-for-scripted-AI-agents), Setup steps, Data Parser node. Checked 2026-09-30.
[^40]: The Sessions view lists sessions and can hide test sessions and filter to sessions where agent handover occurred, an error occurred or a response was downvoted. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), View autonomous AI agent sessions and history section. Checked 2026-09-30.
[^41]: Webex AI Agent Studio is reached from Control Hub at Services > Contact Center > Customer experience > AI Agent, or from Webex Connect through the App Tray menu. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), Access Webex AI Agent Studio section. Checked 2026-09-30.
[^42]: Full Administrators have full AI Agent Studio access including transcripts; Contact Center Service Administrators have full access except transcripts; Read-only Administrators and Supervisors have read-only access without transcripts. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), User roles and access table. Checked 2026-09-30.
[^43]: The Talk to an agent intent's default response template is agent handover, and changing its template key does not change the human-handover outcome. Source: [Understand intents, entities, and responses in AI Agent Studio](https://help.webex.com/en-us/article/sz02k8/Understand-intents,-entities,-and-responses-in-AI-Agent-Studio), Default intents section, Talk to an agent. Checked 2026-09-30.
[^44]: When a scripted agent's Talk to an agent intent is invoked, the system automatically initiates a transfer to a human agent. Source: [Understand intents, entities, and responses in AI Agent Studio](https://help.webex.com/en-us/article/sz02k8/Understand-intents,-entities,-and-responses-in-AI-Agent-Studio), Default intents section, Talk to an agent. Checked 2026-09-30.
[^45]: AI Agent Studio provides autonomous templates Doctor's Appointment, Track Package, Cumulus Bank and Cumulus Airline, and scripted templates Doctor's Appointment and Track Package. Source: [Use AI agent templates](https://help.webex.com/en-us/article/n8mo4c/Use-AI-agent-templates), Available templates section. Checked 2026-09-30.
[^46]: Template agents do not work on their own; fulfillment flows for their actions must be created in the linked Webex Connect tenant before deployment. Source: [Use AI agent templates](https://help.webex.com/en-us/article/n8mo4c/Use-AI-agent-templates), Caveats / next steps section. Checked 2026-09-30.
[^47]: Virtual agent transfer summaries are enabled in Control Hub at Services > Contact Center > Desktop Experience > AI Features, under Generated Summaries, for all queues or selected queues. Source: [Enable AI Assistant features in Control Hub for Webex Contact Center](https://help.webex.com/en-us/article/ndezdu5/Enable-AI-features-in-Control-Hub-for-Webex-Contact-Center), Generated Summaries section. Checked 2026-09-30.
[^48]: AI-generated summaries require the A-FLEX-AI-ASST AI Assistant add-on, and users need AI Assistant access in their Desktop Experience settings. Source: [Enable AI Assistant features in Control Hub for Webex Contact Center](https://help.webex.com/en-us/article/ndezdu5/Enable-AI-features-in-Control-Hub-for-Webex-Contact-Center), Prerequisites section. Checked 2026-09-30.
[^49]: The virtual agent transfer summary appears in the AI Assistant widget on Agent Desktop; the global layout includes it, and custom layouts must add the ai-assistant component to the advancedHeader. Source: [Enable AI Assistant features in Control Hub for Webex Contact Center](https://help.webex.com/en-us/article/ndezdu5/Enable-AI-features-in-Control-Hub-for-Webex-Contact-Center), Generated Summaries section, desktop layout note. Checked 2026-09-30.
[^50]: In the Virtual Agent V2 activity, the administrator selects Webex AI Agent Scripted or Webex AI Agent Autonomous and then picks a published agent from the Virtual Agent list. Source: [Use AI agents for customer interactions](https://help.webex.com/en-us/article/s0qro1/Use-AI-agents-for-customer-interactions), Configure AI agent for voice channel section. Checked 2026-09-30.
[^51]: VirtualAgentV2.ErrorCode reports values including no_error, max_no_input and system_error. Source: [Build and manage flows with Flow Designer](https://help.webex.com/en-us/article/nhovcy4/Build-and-manage-flows-with-Flow-Designer), Virtual Agent V2 activity, output variables. Checked 2026-09-30.
[^52]: If language and voice overrides are not set on the Virtual Agent V2 activity, it takes the input language and output voice from the Global_Language and Global_VoiceName variables. Source: [Use AI agents for customer interactions](https://help.webex.com/en-us/article/s0qro1/Use-AI-agents-for-customer-interactions), Configure AI agent for voice channel section. Checked 2026-09-30.
[^53]: VirtualAgentV2.MetaData returns JSON that includes fulfillment actions, the escalation type and transfer triggers, and VirtualAgentV2.TranscriptURL stores a URL to the conversation transcript. Source: [Use AI agents for customer interactions](https://help.webex.com/en-us/article/s0qro1/Use-AI-agents-for-customer-interactions), Output variables table. Checked 2026-09-30.
[^54]: The Virtual Agent V2 activity has Handled, Escalated and Errored output paths, for a completed self-service interaction, a required escalation to a human agent, and an error. Source: [Use AI agents for customer interactions](https://help.webex.com/en-us/article/s0qro1/Use-AI-agents-for-customer-interactions), Configure AI agent for voice channel section, output paths. Checked 2026-09-30.
[^55]: Cisco says not to place a Virtual Agent V2 activity after a Queue Contact activity or in the Event Flow canvas. Source: [Configure Virtual Agent-Voice in Webex Contact Center](https://help.webex.com/en-us/article/n6gaghu/Configure-Virtual-Agent-Voice-in-Webex-Contact-Center), Limitations section. Checked 2026-09-30.
[^56]: On voice, the Virtual Agent V2 activity's State Event settings (event name and event data) send custom events to the AI agent, and VirtualAgentV2.StateEventName holds a custom event name received from the agent. Source: [Use AI agents for customer interactions](https://help.webex.com/en-us/article/s0qro1/Use-AI-agents-for-customer-interactions), Custom events section. Checked 2026-09-30.
[^57]: Voice preview of an AI agent is available only when AI Agent Studio is opened through Control Hub. Source: [Webex AI Agent Studio Administration guide](https://help.webex.com/article/ncs9r37/Webex-AI-Agent-Studio-Administration-guide), Access Webex AI Agent Studio section, note. Checked 2026-09-30.
[^58]: An entry point is the association of an inbound telephone number with an IVR service flow in Webex Contact Center. Source: [Set up voice channels for Webex Contact Center](https://help.webex.com/en-us/article/nhy3codb/Set-Up-Voice-Channel-for-Webex-Contact-Center), Entry points section. Checked 2026-09-30.
[^59]: A queue receives contacts when a flow's Queue Contact activity selects it. Source: [Understand Routing and Queueing in Webex Contact Center](https://help.webex.com/en-us/article/np2fdx/Understand-Routing-and-Queueing-in-Webex-Contact-Center), Queues section. Checked 2026-09-30.
[^60]: A flow must be validated and its errors fixed before it can be published. Source: [Build and manage flows with Flow Designer](https://help.webex.com/en-us/article/nhovcy4/Build-and-manage-flows-with-Flow-Designer), Publish a flow. Checked 2026-09-30.
[^61]: Entry points are created as channels in Control Hub under Services > Contact Center > Customer experience > Channels by clicking Create. Source: [Set up a channel](https://help.webex.com/en-us/article/ewuay1/Set-up-a-channel), Set up a channel procedure, steps 1 to 3. Checked 2026-09-30.
[^62]: An inbound telephony entry point is assigned its routing flow from a drop-down list in the entry point settings. Source: [Set up a channel](https://help.webex.com/en-us/article/ewuay1/Set-up-a-channel), Entry Point Settings, Routing flow. Checked 2026-09-30.
[^63]: The entry point settings include a version label drop-down used to choose the preferred version of the assigned flow. Source: [Set up a channel](https://help.webex.com/en-us/article/ewuay1/Set-up-a-channel), Entry Point Settings, Version label. Checked 2026-09-30.
[^64]: When creating an inbound telephony entry point, the administrator selects a Webex Calling location and a PSTN number from drop-down lists before saving. Source: [Set up voice channels for Webex Contact Center](https://help.webex.com/en-us/article/nhy3codb/Set-Up-Voice-Channel-for-Webex-Contact-Center), Create entry points procedure. Checked 2026-09-30.