> For the complete documentation index, see [llms.txt](https://docs.eesel.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.eesel.ai/reports/reports.md). # Reports AI support analytics for your agent: how much it handled, how good its answers were, where it got stuck, and written reports on demand or on a schedule. Your AI teammate keeps its own numbers: how much it handled, how good the answers were, where it got stuck. Those numbers tell you which knowledge is missing, which instructions need a fix, and when it is ready to handle more on its own. Ask it in the chat. It reads its own figures, explains them, and can write a fuller report on the spot. Or read the charts yourself on the **Reports** page in the left sidebar.
The Reports page showing time range options, a Total number of tasks chart, and Total trigger events by type broken down by trigger

The top of the Reports page: how much work came in, and which triggers brought it.

## Ask your agent for a report {% stepper %} {% step %} ### Ask in chat Open the chat in your dashboard and ask for the report you want, for example a performance review for the month. Your agent reads its own setup and activity, picks the right report [skill](/skills/skills.md), and saves the report as a file in your agent's **Files**. In the chat it gives you the short version and offers to act on it.
The dashboard chat where the user asks for a performance review report for the month, and the agent gathers its activity data, links the finished report and lists the three changes that would make the biggest difference

The chat after a report ask: the linked review and the three changes it recommends.

**Check it worked.** The report is listed under **Files** in the left sidebar. From a terminal, `eesel files` lists it. {% endstep %} {% step %} ### Open the report Click the link in the chat, or the file under **Files**. The report opens as a document beside the chat, with the numbers first.
The performance review report open as a document, with a summary of the conversations reviewed and an Activity Health Dashboard table of CSAT, resolution rate and knowledge gap rate

The health table at the top of the report.

Below the numbers come the recommendations, each with the evidence behind it and the change to make.
The report's Top Recommendations section, listing the help articles to add for the most common unresolved customer questions with the evidence for each

The Top Recommendations section of the report.

It ends with a summary table you can hand to your team.
The report's Summary table listing four prioritised actions with the estimated number of conversations each would improve

Four prioritised actions, each with the conversations it would improve.

{% endstep %} {% step %} ### Put it on a schedule Ask for a report every week, here a weekly review of your agent's own results, and your agent creates a [schedule](/automations-and-schedules/automations-and-schedules.md), runs it once straight away, and tells you what it found.
The dashboard chat where the user asks the agent to review its setup and results and repeat it weekly, and the agent runs the analysis and sets up the weekly automation

The weekly review asked for in chat, with the analysis running.

The schedule appears on your **Automations** page, switched on. Open it to change the day, the time or what the report covers.
The schedule editor for the weekly agent self-review, set to every Monday at 8:00 AM UTC, with the instructions below

The weekly review schedule, set to Monday at 8:00 AM UTC.

{% endstep %} {% endstepper %} The skills behind these reports: | Skill | What it produces | | ------------------------------- | ----------------------------------------------------------------------------------------------------------- | | **Analyze ticket trends** | The themes that keep coming up across your tickets, with sentiment, volume and how they get resolved | | **Simulation** | A scored report of how your agent handles a batch of your real tickets, with the gaps and the fixes | | **Analyze and improve replies** | The patterns in what your team rejected or edited, and the instruction changes that stop it happening again | | **Review agent setup** | An audit of how your agent is set up and performing, with what to change | ## The Reports page Open **Reports** in the left sidebar. Pick a window at the top: the last 24 hours, 7, 30 or 90 days, or a custom range. | Area | Card | What it tells you | | ------------------ | ----------------------------------- | --------------------------------------------------------- | | **Volume** | Total number of tasks | How much work your agent picked up over the window | | | Total trigger events by type | How that work reached it, by trigger | | **Quality** | AI CSAT distribution | A quality score for the agent's answers across the window | | | AI CSAT over time | The same score day by day | | **Knowledge gaps** | Knowledge gaps | How often it did not have the knowledge to answer | | | Knowledge gaps over time | Answered tasks against flagged gaps, day by day | | **Approvals** | Approval / rejection usage per tool | For each action: approved, rejected, or still waiting | | | Approval efficiency | How much went through first time | | | Review times | How long your team takes to approve or reject | | | Approval trend over time | Approvals and rejections day by day |
Reports cards showing AI CSAT distribution with a positive score, Knowledge gaps split between answered tasks and flagged gaps with a gap rate, and both plotted over time

Quality and knowledge gaps, as a snapshot and day by day.

How to read it: * **The trigger breakdown** shows which triggers bring the most work, and which bring none. * **AI CSAT is not customer feedback.** It is an AI's rating of your agent's answers. Useful as an internal quality signal, not a substitute for asking your customers. * **A knowledge gap is your agent telling you something is missing** from your documentation. Rising gaps on one topic show which articles to write, and the **Update knowledge base** [skill](/skills/skills.md) can draft them. * **Approvals tell you when to hand over more.** An action approved every time without edits is ready to run on its own. One that keeps getting rejected needs an [instructions](/instructions-and-memory/instructions-and-memory.md) fix. Long review times mean approvals are sitting unanswered. ## Activity: the individual runs Reports show the pattern. **Activity** in the left sidebar is where you look at one piece of work. Every task your agent has run is listed there, from every app: a Zendesk ticket, a Slack question, an email and a scheduled run all appear in one history. For each one you can see: * **Where it happened**, with a link to the original ticket or thread * **What set it off**, the trigger or automation * **What it read**, the sources it searched * **What it did**, every action, and where a person approved it * **Why**, its reasoning step by step * **Whether it hit a knowledge gap** When an answer is wrong, read what your agent searched, found and concluded, then fix the cause. Or point your agent at the run in chat. It walks you through its reasoning, offers the fix, and saves the rule to its [instructions](/instructions-and-memory/instructions-and-memory.md) when you agree. ## Exporting Task data can be exported for your own analysis. Exports are prepared in the background and emailed to you as a download link when they are ready. ## FAQ
What does the Reports page show about my AI agent? It shows volume, quality, knowledge gaps and approvals over the window you pick: how many tasks your agent picked up and from which triggers, an AI CSAT score, how often it lacked the knowledge to answer, and how your team handled its approvals.
Is AI CSAT the same as customer satisfaction? No. AI CSAT is an AI's rating of your agent's answers, not feedback from your customers. It is useful as an internal quality signal, not a substitute for asking your customers.
What is a knowledge gap? A knowledge gap is your agent telling you it did not have the knowledge to answer. Rising gaps on one topic show which articles to write, and the **Update knowledge base** skill can draft them.
Can I get a report on my AI agent every week? Yes. Ask for it in chat, and your agent creates a schedule, runs it once straight away, and tells you what it found. The schedule appears on your **Automations** page, where you can change the day, the time or what it covers.
How do I see why my agent answered the way it did? Open the run in **Activity**. It shows where the task happened, what set it off, what your agent read, what it did, its reasoning step by step, and whether it hit a knowledge gap.
Can I export my agent's data? Yes. Task data can be exported for your own analysis. Exports are prepared in the background and emailed to you as a download link.
How do I know when to let my agent act on its own? Watch the approval cards. An action approved every time without edits is ready to run on its own. One that keeps getting rejected needs an instructions fix.
## Related pages * [Skills](/skills/skills.md) for the reporting skills and how to run them * [Automations and Schedules](/automations-and-schedules/automations-and-schedules.md) to have reports arrive on a schedule * [Instructions and Memory](/instructions-and-memory/instructions-and-memory.md) for acting on what the reports tell you * [Use eesel from Claude and other AI assistants](/apis-and-developer-resources/use-eesel-from-claude-and-other-assistants.md) and [Use eesel from ChatGPT](/apis-and-developer-resources/use-eesel-from-chatgpt.md) to ask your agent how it has been doing from a chat assistant --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://docs.eesel.ai/reports/reports.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.