AI Center: Analyze

The AI Center Analyze page is your go-to hub for seeing how AI Copilot and Robin boost your team's efficiency and elevate the end-user experience.

Here, you can track ticket activity for a selected time range, measure the hours saved by Robin, and spot trends showing total tickets alongside those assisted and solved by Robin.

Note:

For your AI optimization level and the guided tasks to improve it, go to AI Center > Setup. See AI Center: Setup.
Full admin permissions are needed to access the AI Center.

Performance stats

View key Robin performance metrics for a selected time range. Click the dropdown to select 7, 30, or 90 days.

Note: The selected time frame applies to all graphs on the Analyze page, and the data in each graph will adjust accordingly.

Autonomous resolution: Percentage of tickets opened and resolved end-to-end by Robin.
Reopen rate: Percentage of tickets Robin marked as resolved that were later reopened.
User satisfaction: Average user rating after a conversation handled by Robin (1-5).
Total hours saved: Estimated time Robin saved your team by handling tickets, measured in hours.

Tickets flow breakdown

Visualize your team's success with a graph showing the full journey of tickets handled by Robin. From intake channel, through Robin's diagnostic and action steps, to the final resolution outcome. It's a quick way to see where volume is concentrated and how effectively Robin is closing tickets end-to-end.

Assigned to Robin: Total number of tickets Robin received and handled during the selected time range, broken down by intake channel.
Actions and diagnostics: Tickets where Robin ran at least one action or diagnostic before reaching a resolution outcome.
No actions and diagnostics: Tickets where Robin reached a resolution outcome without running any actions or diagnostics.
Autonomously resolved: Tickets Robin resolved end-to-end, without technician involvement.
Assisted escalation: Tickets Robin handled but escalated to a technician to complete.

Actions and diagnostics

Track how Robin handled tickets with a detailed breakdown of every action and diagnostic type it ran during the selected time range.

Note: A single ticket can include multiple actions, so the total count may exceed the number of tickets assigned to Robin.

Work out what to improve

Robin resolves tickets two different ways: by knowing something, and by doing something. The Actions and diagnostics panel separates them, which is what makes it useful for deciding where to spend your next effort.

Category	What it tells you
KB article	Robin answered from your documentation. This is your knowledge contribution.
On device diagnostics, On device actions, Cloud Checks, Cloud actions	Robin read or changed something in a real system. This is your action contribution.
Playbooks	Robin followed a workflow you defined, rather than choosing its own steps.

The Tickets flow breakdown gives you a second reading. Tickets that reached an outcome under No actions and diagnostics were resolved by conversation and knowledge alone, with nothing touched on a device.

Reading the split

What you see	What it suggests	Where to spend effort
KB article high, device and cloud categories low	Robin is answering questions well but rarely fixing anything.	Enable more actions, and check that users have completed phone verification. Both produce this pattern.
Device and cloud categories high, KB article near zero	Robin is acting but falling back on general IT knowledge rather than your process.	Add knowledge base articles. Start from Actionable insights, which tells you which questions had no article behind them.
Playbooks at zero while Assisted escalation is high	Robin handles what it can generically and hands back everything that needed your specific process.	Build Playbooks for your most common escalations.
Both sides active but Autonomous resolution still low	Robin is doing work without finishing it.	Look at the escalation path rather than at either contribution. The gap is usually a missing final step, not missing knowledge or actions.
Reopen rate rising	Whichever side is growing is resolving things that were not actually fixed.	Treat this before adding anything. A rising reopen rate makes both contributions look better than they are.

Two things to keep in mind

Note: One ticket can include several actions, so the category counts add up to more than the number of tickets. Read them as shares of activity, not as ticket counts.

Note: A category at zero does not mean the capability is missing. It means Robin has not used it in the selected time range, which may be because it is not enabled, not because it does not exist.

Tip: Change the time range and look again. A category that appears over 90 days but not over 7 is being used occasionally rather than not at all, and that is a different problem from one that has never fired.

Hours saved on tickets

Robin works around the clock to save your team valuable time every day. The graph visualizes this, showing a daily breakdown of total hours saved by Robin alongside the calculated daily average.

Hover over any date to view details of where those hours came from, including how many tickets Robin assisted with and how many it resolved that day.

How are the hours saved calculated?

The hours saved are estimated by AI, comparing technician actions to those handled by Robin. For each ticket, the human-equivalent effort that Robin eliminates is calculated by analyzing the actions performed, measuring how long those actions would typically take for a human technician, and subtracting the time they took Robin to execute. The result is an estimate of technician time saved added to the total time saved by Robin.

These examples of real ticket conversations highlight how Robin saved time by resolving issues and assisting users, showing exactly how much time was saved at each step.

Example 1: Solved by Robin

A user reported they couldn't connect to the corporate VPN.

Robin ran diagnostics on the VPN connection, detecting a VPN adapter issue before guiding the user through a repair attempt, and then a full automatic reinstallation of the VPN client.

Time-saving actions taken by Robin:

Detected VPN adapter issue and proposed a restart (4 min): Realistically, a technician would need a few minutes to analyze such a connectivity problem and suggest a restart as a quick remedy.
Guided the user through a VPN client repair attempt (6 min): A technician would typically have to spend several minutes finding the information and explaining each step in detail, especially if it involved multiple interface interactions expecting user confirmation.
Suggested and initiated FortiClient reinstallation (8 min): This would take around 8 minutes for a technician as it involves assessing reinstallation needs, executing, and monitoring the installation progress.

In total, Robin saved 18 minutes, time that would typically require a technician to troubleshoot and fix manually, resolving the issue instantly and allowing the user to reconnect to the corporate network.

Example 2: Assisted by Robin

The user reported needing access to a specific file in the company platform database.

Robin first checked internally for any required permissions or guidance, reviewing the organization's knowledge base to find the proper process. It identified that approval from the Finance team was needed and provided the user with the necessary instructions. Then, without waiting for further input, Robin escalated the request to a technician to complete the final step. By handling everything up to the approval, Robin minimized back-and-forth communication and sped up the process.