Conversation Closure: Handover (N2) and Retain (N1)

Descriptions of our handover functionality and ticket closure

By Fabrício Rissetto and 4 others

• 11 articles

N2 | Notify that there are no available people in N2 outside of business hours (Humans office hours)

This feature allows us to configure the hours that the N2 support team is available, so that ClaudIA can adjust the handover message accordingly. The scheduling of hours follows the Cron Job Schedules pattern (PS: we know this is not something very simple to use, we intend to improve this in the future). Fields to be configured by you: - Schedule (use the website https://crontab.guru/ to automatically generate the Cron strings for you) - Available (checkbox) More details in the video below: https://www.loom.com/share/47e49cb0a2b4423e827e202c6b90d774 How it works Each "Schedule" field in the print below defines a time at which ClaudIA should be updated according to the checkbox on the right, "Available". When that time is reached, ClaudIA updates the status of humans to available or unavailable according to the checkbox on the right of each Cron. This availability of humans affects other behaviors of ClaudIA, such as the use (or not) of the handover message configured in another field on the same screen:

Last updated on Aug 12, 2025

N2 | Remove Message in Handover

The Handover Without Sending Message feature was developed to allow the escalation of tickets without sending messages notifying the customer, recording only an internal note in the Helpdesk. This improvement avoids unnecessary communications and ensures that the team can manage tickets more efficiently, maintaining internal transparency without impacting the end-user experience. NOTE: This feature is disabled by default in projects; to enable it, you need to contact us to request it. IMPORTANT: It cannot be activated together with the "DETECT TRANSFER" feature, where we perform a check for sending messages with transfer words of the service.

Last updated on Aug 12, 2025

N1 | Ticket Closure (Ticket Resolver)

Ticket Closure Settings (Ticket Resolver) The Ticket Resolver allows you to set rules for the automatic closure of tickets after ClaudIA's response, followed by a period of user inactivity. This ensures that unresolved tickets do not remain open unnecessarily and that the customer receives feedback that the conversation is being closed. Adjusting Settings in the Hub In the Hub settings screen, you can customize these options: - "Minutes until ticket is resolved": sets the time, in minutes, required to automatically close the ticket after the user's last interaction. - Ticket closure messages: There are two types of messages that can be configured when closing a ticket: 1. Without feedback request: message sent when no feedback is requested from the user, such as resolution feedback or a CSAT rating. 2. With feedback request: message sent when the user is asked to provide feedback or a rating before the ticket is closed. These messages can be independently customized, allowing adjustments according to the desired flow. In many cases, both can have the same content, but if necessary, they can differ for specific situations. With these settings, you can optimize the service experience and ensure an efficient automatic closure process.

Last updated on Aug 12, 2025

N2 | Handover when there is instability (Handover by timeout)

When does the timeout handover occur? This behavior occurs when there is some instability that prevents returning to the customer within the expected time. In these cases, the functionality automatically performs the handover of the ticket to ensure the continuity of customer service. Since it is an automatic action and anticipated by the functionality, this type of handover is not available for auditing. In the ticket, there will be an indication in the internal note:

Last updated on Aug 12, 2025

N2 | Transfer due to lack of content (Low Confidence & Count Limit)

One of the controls we have to identify that ClaudIA is not able to find enough content to respond to the user is the 'low confidence counter'. In the Hub, when ClaudIA does not find content above the stipulated confidence (more on that later), a 'Count limit' indicator appears in yellow below ClaudIA's response, as shown below: After reaching the configured limit, the service is automatically transferred to a human using the message configured in the 'Forward To Human Message' field on the Hub settings screen. The configuration of the limit counters can also be adjusted on the same screen: - 'Limit' is the field that defines the maximum number of low-confidence searches. PS: we recommend at least 3, as not every low-confidence content search necessarily generates a bad response. Sometimes ClaudIA clarifies and that resolves it; sometimes the user's message does not require content to be answered (example: "Good morning!") - 'Low Confidence Threshold' is the minimum expected distance to be considered relevant content (high confidence). Every content search in the Hub presents a score that you can track directly whenever you see the sessions returned by ClaudIA's search, as in the Conversations audit screen: Additional functionality It is possible to configure ClaudIA to ignore the low confidence count on the first response of a ticket, considering that often the user's first interaction is just a greeting. This prevents the counter from being incremented unnecessarily. However, if the first message already contains a question and ClaudIA provides a low-confidence answer, that interaction will not be counted. The other interactions in the ticket follow the normal confidence counting flow. Currently, this configuration is not available in the Hub. To activate it, you need to request a Clouder to set the option ignoreConfidenceCountOnFirstMessage to true.

Last updated on Aug 12, 2025

N1 | Validate problem resolution before closing (Resolved Issue Verifier)

How it works 1. Final question triggered by the timer (default flow) - ClaudIA sends a follow-up message (e.g., "Can I assist you with anything else?") - The system interprets the user's response: - Resolved: ticket is automatically closed after a few minutes (asynchronous routine). - Dissatisfaction: ticket is moved to N2. - Other response: the conversation remains open. 🔸 Important: the service runs asynchronously every 5 minutes, so there may be a slight delay in triggering. 2. Verification with each new message (additional option) - ClaudIA proactively evaluates each new message sent by the client. - Examples: - "That’s it, thank you" → immediate closure (without waiting for the timer). - "It’s not resolved" → conversation continues open, possibly moving to N2. - Neutral response (“Ok”, “Thanks”) → ClaudIA responds and sends the closing follow-up. How to set up in the Hub 1. Access the Hub; 2. Go to Settings; 3. Click on the Resolved Issue Verifier tab; 4. Check the option to enable or disable the feature; 5. Set the wait time for the question to be sent (“Can I assist you with anything else?”). 🔸 Important: the service that sends this question runs asynchronously every 5 minutes, so the configured time may have up to a 5-minute delay; 6. Customize the message that ClaudIA will send at this moment; 7. Verify if the problem was resolved on each new message (optional): by activating this feature, ClaudIA automatically evaluates each user message and can close the ticket when it detects that the problem has been resolved. Important: this setting will also use the previously configured "Question" for validation of completion. ⚠️ Interaction with Eddie flows Here are the new features (with feature flag): 1. Send follow-up but keep user in Eddie - When the customer responds to a follow-up during an Eddie flow, the conversation does not return to ClaudIA - Control remains in Eddie until the flow ends. The user's response post-FUP goes to Eddie. 2. Do not send follow-up in Eddie - If the conversation is ongoing in an Eddie flow, ClaudIA does not send the follow-up question. - This way, Eddie continues the conversation without interruptions. However, keep in mind that the ticket may be closed per the default auto-close time set for conversations, without sending a follow-up. These two behaviors are disabled by default but can be activated upon request — contact the support team to enable via feature flag.

Last updated on Oct 02, 2025

N2 | Ticket Taking Functionality in the Hub, Removing It from Service (Forced Handover)

https://www.loom.com/share/d68947c57c0c4675b4ff6474b58830c9?sid=955f0f63-8092-4b3d-b0d2-5f5b410b484b The action of taking a ticket allows your team to interrupt the automated service provided by ClaudIA and take manual control of the conversation with the end customer. When should I use this functionality? You can use this functionality to take over a ticket that is being handled by ClaudIA. This is useful in situations such as: - The automated service is not progressing as expected. - You want to make a more precise manual intervention. - Your team wants to escalate the conversation to a more human or specialized handling. The Take All button is disabled. Why? This button is only enabled when there are open conversations in the current filter, i.e., conversations that have Status: Open, still in progress with ClaudIA. If the button is disabled, it is because: - The current filter shows only resolved or closed conversations. - No eligible conversations were found in the selected period. How to view open conversations? 1. Click on Filter. 2. In Resolution, select the Open or All option. 3. Click on Apply filters. What happens when I click Take? When you click Take (individually or in bulk), the following steps are performed: 1. We attempt to automatically transfer the conversation to Level 2 (N2) of service. 2. Even if the transfer is unsuccessful, ClaudIA interrupts the service of that ticket. 3. From that moment on, your team can take over the service manually, redirect, or end the conversation as preferred. Important: In cases where automatic transfer does not occur, you or someone from your team needs to pull it manually for you. This is done in your own Helpdesk. These cases happen when there is some communication failure, so we prioritize ensuring that ClaudIA leaves the service completely, rather than having the chance to transfer to someone and it responds in the middle of the conversation—which we know brings a very bad experience for the end customer. Button to take tickets in bulk: Button to take a ticket individually:

Last updated on Aug 12, 2025

N2 | Detect Transfer Intent in ClaudIA's Response (Detect Transfer)

What is Detect Transfer and How to Map Detect Transfer Cases? Detect transfer is a mechanism for transferring service to a human agent (N2). The percentage of tickets escalated via detect transfer can be found on the Cloud Humans Hub Dashboard, in the "Retention Metrics" tab, in the "Transfer Reason Proportion in N2 Tickets" graph, labeled "ClaudIA detected transfer action". Tickets can be found in the table below "Number of Tickets/Transfer Reason" under the name "ClaudIA said she would transfer the service". How to Activate Detect Transfer? Detect transfer is activated when ClaudIA sends a message to the user suggesting that the service will be transferred to a human, with phrases like "I will transfer your service to a specialist", "I will forward your request to an agent", "I will transfer your situation to our team". Therefore, for detect transfer to be activated, it is necessary to include phrases that trigger it, like the examples above, in the responses of sections that should escalate the service to a human. Important: phrases that trigger detect transfer SHOULD NOT be included in N1 or INTERACTIVE sections. Otherwise, ClaudIA may improperly transfer services to humans when they should continue with ClaudIA or be forwarded to an Eddie.

Last updated on Aug 12, 2025

N2 | Trying to Convince the User to Talk to AI Before Transferring (Forward to Human Check)

This feature allows ClaudIA to try to convince the user to be served by her instead of being directly transferred to a human when requested. This is especially useful in cases where, before ClaudIA, there was a TREE BOT or any automation alternative that provided a poor experience to users, and they have already developed the habit of requesting human assistance. In many cases, we are able to convert this user to stay with ClaudIA. If the user insists, we respect the request and transfer the ticket to the N2 team. Example of a user insisting to talk to a human To increase the retention rate, we can configure the message to be more persuasive, something like "The team is experiencing high demand and it may take a few minutes to respond..." Example of a user who accepted to be served by the AI Interaction with Eddie flows It is worth noting that, although this feature was created with the goal of increasing retention, it does not "lock" the user. All of ClaudIA's guardrails are designed with a focus on quality — even when optimizing for retention. In other words: if the user insists, they will be directed to a human, as part of our product's design choice. A direct impact of this decision is that, even within a structured flow in Eddie, if the customer asks to speak to a human, they will be taken out of that flow and redirected to human assistance.

Last updated on Aug 12, 2025

N2 | Handover when interaction is too long (Message Limit)

Feature Context In many chat support scenarios, it's common to set limits on the number of messages exchanged between a user and an agent. This limitation can be useful for various reasons, such as preventing another bot from engaging with Claudia and causing the conversation to loop indefinitely, or even handing over to a human to check the reasons why a conversation is extending too long (for security reasons or another reason). The default limit for this configuration across all projects is 30, and it can be changed if necessary upon request to our team. Whenever the number of messages exceeds the value configured in this field, the system returns the internal note: "Handover by message limit exceeded" This message indicates that the defined limit has been reached, and therefore, the conversation will be transferred. Message Counting Logic Whenever the user or the agent sends a message, the counter increments. The total sum of messages (user + agent) is compared with the defined value. When this sum exceeds the value, the handover action is triggered. Behavior when the limit is exceeded Upon exceeding the configured limit, Claudia forwards to a human agent.

Last updated on Aug 12, 2025

N2 | Frustration Detection Transfer

What is it? ClaudIA can detect when a customer becomes frustrated during a conversation. When this happens, she can automatically transfer the support to a human agent, ensuring the issue is handled more carefully. How it works Every message from the customer is analyzed and assigned a frustration score. This score accumulates until it reaches the configured limit. - 1 point — Mild irritation The customer shows impatience or makes repeated complaints, but without aggression. - 3 points — Frustrated Stronger complaints about the service/product, tense tone, risk of cancellation, mention of billing/refund, or occasional use of mild swear words. - 5 points — Highly frustrated The customer threatens to cancel, writes IN ALL CAPS, asks to speak with a supervisor, uses strong swear words frequently, or threatens legal action. When the total reaches 5 points (or the configured value), ClaudIA transfers the support. She sends a standard message to the customer and logs an internal note explaining the reason for the transfer. How to configure In the Hub, under the Settings tab, you can: - Enable or disable the feature. - Set the maximum allowed score (default is 5 points). - Edit the message sent to the customer when frustration is detected. How to identify in the Hub Whenever this feature is enabled, the reason for the transfer will be recorded with “Reason: Frustration detected”. This appears in both the Hub and the Helpdesk.

Last updated on Sep 03, 2025