Greater Traceability of wvx Copilot Responses in Reports

The wvx Copilot detailed report now includes additional information to identify the knowledge domain used and the origin of each response.

The following fields have been added:

• DOMAIN: indicates the knowledge domain used by wvx Copilot to generate the response.
• ANSWER_ORIGIN: identifies the origin of the response, distinguishing between a response generated by Copilot (Copilot) and a spelling correction result (Corrected).

The wvx Copilot history in wolkvox Manager desktop also includes a new column showing the knowledge domain used.

Important: records generated before this update may have empty values in these new fields because this information was not previously stored.

Benefits

• Allows you to identify which knowledge domain was used to handle each query.
• Makes it easier to distinguish between AI-generated responses and spelling corrections.
• Improves traceability and monitoring of wvx Copilot usage within the operation.

New Permission to Access AI Reports

A new permission called "AI Reports" has been added to the profile configuration in wolkvox Manager.

This permission allows you to control which user profiles can access the operation's voice and text analytics reports.

To configure it, go to the "Users and Profiles" section in wolkvox Manager, select the corresponding profile, and enable the "AI Reports" option.

When the permission is disabled, AI analytics reports will not be available to users assigned to that profile, even if the operation has subscribed to the service.

Important: this permission is disabled by default, including for existing profiles. The administrator must enable it for profiles that require access to these reports.

Benefits

• Provides greater control over access to voice and text analytics information.
• Allows you to define which profiles can view the results of AI-powered analyses.
• Facilitates permission management according to each user's responsibilities.

New Security Option to Hide Stored Agent Passwords

The security settings in wolkvox Manager now include the "Hide Stored Agent Password" option.

When this feature is enabled, previously saved passwords are no longer displayed in the agents' configuration information.

When the option is enabled:

• The agent's password field appears empty when viewing their configuration.
• The stored password cannot be viewed or copied directly.
• The administrator can enter and save a new password when necessary.

This setting does not modify existing security policies related to password length or password change requirements established for users.

Benefits

• Reduces exposure of stored credentials.
• Strengthens the protection of agents' login information.
• Allows passwords to continue being managed without displaying existing credentials.

Improvements to Visual Flow Editing in wolkvox Studio Web

New editing tools have been added to the flow design canvas in wolkvox Studio web, making it easier to organize and connect components.

The improvements include:

• Inserting components into existing connections: allows you to drag and drop a component directly onto a connection between two nodes to add it to the flow sequence.
• Connection customization: allows you to select cubic or curved connection styles.
• Component reconnection: allows you to modify a connection by dragging one of its endpoints to another node.
• Label display: allows you to identify the labels corresponding to the different branches of the diagram.
• Component removal: allows you to remove a node from a sequence without interrupting the connection between the remaining components.

Benefits

• Speeds up flow creation and modification.
• Reduces the need to rebuild connections when adding, modifying, or removing components.
• Makes it easier to visually understand the paths and decisions configured in a flow.

Automatic Voice Note Detection in Chat Flows

Chat flows compiled from wolkvox Studio web now automatically enable the detection of voice notes received from customers.

Previously, this functionality depended on explicit activation in the flow's initial configuration. With this improvement, detection is enabled when the flow is compiled.

For example, when a customer sends a voice note through WhatsApp, the flow can identify this type of content without relying on the previous manual activation process.

Important: chat flows published before this version must be compiled and published again to incorporate this change.

Benefits

• Simplifies the configuration of flows that receive voice notes.
• Reduces the need for manual configuration to detect audio messages.
• Facilitates the use of voice messages within conversational experiences.

Real-Time Conversation Analysis with Agentic Analytics

Agentic Analytics introduces real-time analysis capabilities that allow you to access information from voice and chat conversations while they are still active.

When this feature is enabled for the operation, live analysis provides information such as:

• Sentiment: sentiment identified as the conversation progresses.
• Risk: risk-related information detected during the interaction.
• Suggested Code: code proposed by the artificial intelligence analysis.
• Latency: information about response times recorded during the conversation.
• Recognized Text: textual content identified during the interaction.

The information is updated while the conversation remains open, allowing you to monitor its progress without necessarily waiting for it to end.

This functionality requires real-time analysis to be enabled for the operation.

Benefits

• Allows you to monitor conversation behavior while interactions are in progress.
• Facilitates early identification of risk situations or changes in customer sentiment.
• Provides additional information for monitoring AI agent performance.

New Knowledge Sources and Synchronization in Copilot Studio

Copilot Studio expands the available options for managing the knowledge sources used by wvx Copilot.

In addition to uploaded documents, web pages can now be added as information sources for the assistant.

For web sources, it is now possible to configure synchronization intervals and view information about the result of the most recent update attempt.

The new features include:

• Files: management of documents used as knowledge sources.
• Web Pages: incorporation of content from a URL.
• Scheduled Synchronization: configuration of intervals to update content from web sources.
• Synchronization Status: access to the results of update processes, including error information when applicable.

In wolkvox Manager desktop, Copilot Studio is available under Settings > wvx Copilot > Copilot.

Benefits

• Expands the possibilities for incorporating business information into knowledge bases.
• Simplifies the maintenance of content obtained from web pages.
• Allows you to monitor synchronization processes to identify potential update errors.

New CRM Agent Component 

This is an advanced artificial intelligence component integrated into the chat component palette in Wolkvox Studio that enables data management and manipulation in the operation's CRM through natural language instructions. Configured by the designer with the CRM server, authentication token, AI model (Smart or Reasoner), and at least one tool from the internal catalog, the node operates autonomously without requiring traditional code blocks or fixed endpoints. The system loads the platform's actual modules and fields to dynamically plan up to 5 execution cycles (with a limit of 12 steps per plan and 30 tool calls) to resolve complex requests involving data retrieval, creation, updating, or deletion. The synthesized response is consistently stored in the $crm_agent_response variable, allowing the conversational flow to continue immediately through its output connection.

Benefits

• Autonomy and Conversational Flexibility: Performs complex CRM operations by interpreting natural language without requiring the development of integration nodes or rigid APIs.
• Data Consistency and Integrity: Queries the actual module and field structure of the CRM server before taking action, preventing hallucinations or errors in attribute names.
• Contextual Persistence: Provides an optional session memory mechanism that stores a conversation summary to maintain continuity across recurring interactions.

New YouTube Channel Integration

Wolkvox connects your YouTube channel: comments and live chat messages from your videos appear in the same inbox where your team handles other interactions.

The connection is authorized using the Google account that owns the channel. From that moment on, each comment can be assigned, tagged, replied to, and closed, and the response is published on YouTube without leaving Wolkvox.

Benefits

• Unified Inbox: Comments and live chat messages appear alongside interactions from other social networks, with the user's history available for reference.
• Managed Public Responses: Responses are sent from AgentBox with full traceability.
• Unified Reporting: Activity is recorded in Wolkvox reports, just like interactions from other channels.

Instagram Conversation Classification

This enhancement introduces the new $source_conv variable, which clearly identifies the exact entry point of each Instagram interaction.

By automatically distinguishing whether a customer mentioned the brand in a story (story_mention) or replied to a brand story (story_reply), the platform enables service segmentation, personalized user experiences, and differentiated sales automations from the very beginning of the interaction.

Benefits

• Personalized Customer Experience: The flow responds based on the interaction's origin and can stop processing a story mention so that it is not treated as a chat or routed to an agent.
• Contextual Responses: The bot knows which story or promotion prompted the customer to initiate the interaction and responds accordingly, without using a generic greeting.
• Measurable Campaigns: The source_conv variable can be used to count how many conversations originate from each Instagram story or mention.

Agentic Flow

Agentic Flow is a conversational assistant embedded in a side panel within the Wolkvox Studio visual editor. It allows users to design and modify flow diagrams (voice IVR, chat, or CRM) using natural language. Users can request the creation, adjustment, or restructuring of nodes and connections instead of drawing them manually. When an instruction is submitted, the system serializes the current diagram into JSON format and processes the modifications using AI models, returning the new structure to the visual canvas as a draft ready to be reviewed and compiled before publication.

Benefits

• Agile Flow Development: Dramatically accelerates the development and restructuring of operational diagrams by translating text instructions into functional nodes and connections.
• Multi-Mode Operational Flexibility: Provides analytical and informational responses about the diagram without modifying the flow (Ask Mode), or actively edits the graph architecture (Agent Mode).
• Resilience and Data Protection: Implements automatic backup mechanisms that restore the original diagram if the model fails to return an executable or error-free structure.
• Assisted Requirements Gathering: Proactively guides the designer by requesting key steps and exact greetings when attempting to build a diagram from scratch.