To view the usage statistics of your AI agents in Papo (https://papo.global/), access the Prompt and AI module and open the Statistics area. There you will find metrics such as successfully sent messages, failures, and skipped interactions, which show exactly how many automatic responses were delivered and in which situations the AI chose not to respond. This data is essential for adjusting prompts, rules, and the knowledge base, keeping automation aligned with customer expectations.
Continuous analysis of the AI's behavior helps identify failure patterns and calibrate the tone and accuracy of responses. Unlike performance reports for human agents (or the Capitão assistant), the Prompt and AI Module statistics focus on technical delivery: they show whether the message generated by the model reached the customer and why, in some cases, it was not sent.
Summary of metrics available in the statistics
| Metric | What it shows | How to interpret |
|---|---|---|
| Succeeded | Number of messages that the AI agent sent without errors. | Indicates the real volume of completed automated interactions. The higher it is, the more demand the AI is absorbing. |
| Failed | Response attempts that were not delivered due to a technical error or block. | Helps identify integration issues, message limits, or incorrect inbox configuration. |
| Skipped | Situations in which the AI decided not to respond — due to lack of confidence, a defined rule, or understanding that it should not intervene. | Reveals gaps in the knowledge base or prompt restrictions. A high number of skipped messages can mean that customers are asking questions for which the AI was not trained. |
How to access Prompt and AI Module statistics
The metrics are concentrated within the Prompt and AI module itself, with no need to navigate to general reports. The path is direct:
- Enter the Prompt and AI module – In Papo's side menu, click Prompt and AI. If the item does not appear, your administrator may have restricted access with custom user roles; ask them to review permissions.
- Open the Statistics view – Within the module, find the Statistics option (in a tab or submenu, depending on the interface version). It is the panel that gathers the AI's charts and send counts.
- Apply the period filter – Use the time shortcuts (today, last 7 days, 30 days) or set a custom range. This way, you can, for example, check the impact of a recent prompt change.
- Analyze the three main metrics – The panel displays the counters for Succeeded, Failed, and Skipped. A visual summary usually shows the proportion between these categories, helping you quickly see whether the AI is performing well or needs adjustments.
- Export if necessary – Although the main purpose is quick consultation, Papo allows you to capture the data for external analysis through operational reports or the API, if you want to cross-reference the numbers with other sources.
Who benefits most from this monitoring
- Automation and support analysts who adjust prompts daily and need to measure the effect of each change.
- Customer service managers who want to quantify how much load the AI is actually absorbing — comparing the number of successes with the total conversations handled by the human team.
- White label resellers who manage multiple client accounts and need to prove the effectiveness of the configured AI, offering data-driven consulting.
- Operations teams who investigate spikes in failures — for example, when a channel integration becomes unstable and generates errors in automatic responses.
Frequently asked questions
What do the three main metrics mean?
Succeeded are the messages that the AI generated and the system managed to send to the customer. Failed are attempts that resulted in an error (channel unavailable, expired token, spam blocking, etc.). Skipped represents cases in which the AI, following its rules or confidence level, chose not to respond — for example, when the customer's question did not fit the scope defined in the prompt.
Do the Prompt and AI Module statistics show the performance of all agents?
Yes. The metrics are consolidated per account, reflecting all agents (the "General" one and any sub-assistants) that were active in any inbox during the period. If you want to analyze a specific agent, you can cross-reference this data with the list of inboxes and the agent reports available on the platform.
Are these statistics updated in real time?
The metrics are updated automatically as interactions occur, allowing you to monitor the AI's behavior almost in real time. The dashboard reflects the most recent interactions processed by the system.
How to interpret a high number of skipped messages (Skipped)?
A high volume of "Skipped" indicates that many conversations did not receive an automatic response because the AI was not confident or because a rule prevented it. Review the prompt, expand the knowledge base, or adjust the general rules so the agent covers more scenarios. In some cases, it can also mean that customers are asking very open-ended questions — consider including example responses in the prompt.
Do these statistics replace the Agent Report?
No. The Agent Report (Reports module) shows performance metrics such as CSAT, average response time, and resolved conversations, including those of the Capitão assistant. The Prompt and AI Module statistics focus on the technical delivery of messages by automation. The two views complement each other: one measures satisfaction and efficiency, the other measures the technical robustness of delivery.
Where can I see the performance of the Capitão assistant (team copilot)?
Capitão is an internal AI agent that assists your team during customer service. Its statistics (resolved conversations, CSAT, SLA) are in the Agent Report, accessible through the Reports module. The Prompt and AI Module statistics do not include these indicators, as they focus on messages automatically sent to customers by the agents configured in Prompt.
Configure your AI agents in minutes and view usage statistics directly in the Papo dashboard.