Agent Logs

Agent Logs gives you insight into how your users interact with Luzmo Agents, both in-app and in embedded experiences. Review what users ask, which answers they receive, which data sources and AI context were used, and whether users gave positive or negative feedback.

Benefits

• Understand usage and adoption: see how users interact with agents and which questions they ask. Identify common requests to better understand their needs.
• Review answers: open individual interactions to inspect the prompt and generated text, charts or dashboards. This helps you investigate unexpected results and reported issues.
• Improve AI context: check which datasets and field descriptions informed an answer, and spot missing context. Use these insights to clarify terminology or business definitions that agents need to interpret your data correctly.
• Learn from user feedback: see which results received positive or negative feedback. Use negative feedback to identify interactions that need closer review and positive feedback to understand which answers users find helpful.

Access and activation

Only Organization Owners can access the page.

To start collecting logs:

1. Go to Settings → Agent Logs.
2. Review the data collection notice.
3. Click Activate Agent Logs.

No Agent Logs data is collected before you opt in. Earlier interactions will not appear in the logs.

By activating Agent Logs, you authorize Luzmo to process this data on your behalf. You are responsible for notifying your end users.

Data collection and retention policy

Agent Logs stores user prompts and generated results so you can review agent interactions within your organization.

• 30-day retention: Agent Logs data is retained for 30 days, then automatically and fully deleted.
• No model training: This data is never used to train models or for analytics beyond providing you with these logs.
• Disabling logging: Contact Luzmo via support@luzmo.com if you want to stop collection. When logging is disabled, existing Agent Logs data is also deleted.

How to use Agent Logs

In Luzmo, go to Settings → Agent Logs to view your logs.

The summary cards at the top of the page show:

• Prompts logged: the number of logged prompts and generated results.
• Feedback coverage: how many prompts have received feedback and how many remain unrated.
• Flagged as wrong: negative feedback that may need review.
• Positive signals: positive feedback on agent results.

The table displays one row per question and its generated results. Each row includes the timestamp, prompt, source, feature, user, result types and feedback.

Use Expand filters to narrow down the logs. Click any row in the table to open the Agent result detail view, where you can review the full prompt, generated results, data context and feedback.

Reviewing an agent result

The details page shows the user's prompt alongside the generated text, charts or dashboards.

You can also review:

• User details: information about the user who submitted the prompt.
• Data context used: the datasets and AI context used to generate the result, including fields where no AI context is set.
• More information: the source, feature, feedback and response time.

Use this information to investigate an unexpected answer. For example, if an agent misinterprets a field, check whether that field has AI context explaining its meaning.

Chart and dashboard previews

Chart and dashboard previews apply the original filters to current data. If the underlying data has changed, the preview may differ from what the user originally saw.

Keep this in mind when comparing a preview with the logged text answer or investigating user feedback.