AI Insights is an optional feature of Morbit Studio that turns the data in a report into a short, plain-English analysis: a summary of what the numbers say, the risks and concerns they point to, and a ranked list of recommended actions. This guide explains what AI Insights does, which reports support it, exactly what data is analysed, and the security and privacy controls around it.
Contents
- What is AI Insights?
- How to generate an AI Insight
- What the output contains
- Reports that support AI Insights
- What data is sent to the AI
- Security and privacy
- How we keep the output consistent
- Getting the best results
- Frequently asked questions
What is AI Insights?
Morbit Studio collects a large amount of operational data: call records, device registrations, peripheral usage, room bookings, occupancy and environmental sensor readings. Reading through that data to work out what matters takes time and experience. AI Insights does the first pass for you.
When you run AI Insights on a report, the data currently shown in that report is analysed by a large language model and returned as a structured briefing written for business, operations and support stakeholders. The aim is not to replace the report, but to answer the question “what should I do about this?” quickly, so you can spend your time acting rather than interpreting.
How to generate an AI Insight
- In Morbit Studio, open one of the supported reports.
- Set the report filters as you normally would, for example the date range, organisation, site or group. AI Insights analyses the report exactly as filtered, so narrowing the scope gives you a more focused result.
- Click the AI Insights button on the report page.
- Wait a few moments while the analysis runs. Larger reports take a little longer.
- Read the analysis. It is generated fresh each time; run it again if you change the filters.
Each AI Insight is produced on demand for the report you are viewing. Nothing is queued, scheduled or emailed automatically.
What the output contains
Every AI Insight follows the same broad structure so that stakeholders can read it quickly and compare one report with another:
| Section | What it tells you |
|---|---|
| Summary | The headline figures from the report, for example the percentage of devices online, the average booking rate or the proportion of calls rated as good quality, with a short explanation of what they mean in context. |
| Risks and concerns | Patterns in the data that need attention: rooms that are booked but not used, devices with repeated quality problems, sites that lag behind the rest of the estate, support cover that is due to expire, and so on. |
| Recommended actions | A list of actions ranked by priority. Each action includes a suggested owner (for example the AV team, facilities or the service desk) and a suggested timeframe, so the output can be dropped straight into a task list or review meeting. |
The recommendations are advisory. They are based only on the data in the report and should be checked against what you know about your environment before you act on them.
Reports that support AI Insights
AI Insights is available on the following reports. Descriptions of each report are in the Morbit Studio Administrator Guide. Zoom Room and Zoom User reports follow the same principles as the equivalent MS Teams reports.
Call reports
| Report | What AI Insights looks at |
|---|---|
| Call List | Individual inbound and outbound calls, their type, duration and quality rating. |
| Call Usage | Call volumes and durations over the selected period. |
| Call Distribution | How calls are spread across sites, devices and times. |
| Call Protocols | Which call protocols are in use and in what proportion. |
| Call Quality | The ratio of good, OK and bad calls and where the poor calls are concentrated. |
| Group Participants | Participation in group calls and conferences. |
MS Teams reports
| Report | What AI Insights looks at |
|---|---|
| Registration Usage | How much each Teams device or registration is being used. |
| Registration Quality | Call quality by Teams registration. |
| Registration Activity | Device activity and status changes over time. |
| Peripheral List | Connected cameras, speakers and microphones, and how often each is used. |
| Peripheral Overview | A summary of the peripheral estate across all devices. |
| Operating Systems | Operating system versions in use across Teams clients and devices. |
Zoom Room reports
| Report | What AI Insights looks at |
|---|---|
| Room Usage | How much each Zoom Room is being used. |
| Room Quality | Call quality by Zoom Room. |
| Room Activity | Room activity and status changes over time. |
| Room Peripheral List | Peripherals connected to Zoom Rooms and how often each is used. |
| Room Peripheral Overview | A summary of the Zoom Room peripheral estate. |
| Room Operating Systems | Operating system versions in use across Zoom Rooms. |
Zoom User reports
| Report | What AI Insights looks at |
|---|---|
| User Usage | How much each Zoom user is using the platform. |
| User Quality | Call quality by Zoom user. |
| User Activity | User activity over time. |
| User Peripheral List | Peripherals used by Zoom users and how often each is used. |
| User Peripheral Overview | A summary of peripherals across Zoom users. |
| Zoom User Operating Systems | Operating system versions in use across Zoom user clients. |
Environmental reports
| Report | What AI Insights looks at |
|---|---|
| Room Sensors | Readings from the environmental sensors in each room. |
| Room Comfort | Temperature, humidity and CO2 against comfort thresholds. |
| Room Violations | When and where environmental conditions fell outside the acceptable range. |
Room usage reports
| Report | What AI Insights looks at |
|---|---|
| Room Bookings | Booking volumes and occupancy patterns by room. |
| No Shows by Room | Rooms that are booked but not occupied. |
| No Shows by Time | When no-shows happen during the day and week. |
| No Shows by Organiser | Which organisers most often book rooms that go unused. |
Desk usage reports
| Report | What AI Insights looks at |
|---|---|
| Desk Utilization | How desks are used across the estate and where capacity is over or under-used. |
SIP/H.323 reports
| Report | What AI Insights looks at |
|---|---|
| Device Usage | Activity on SIP and H.323 endpoints such as Cisco and Poly devices. |
| Device Quality | Call quality for SIP and H.323 endpoints. |
| Active Devices | Which devices are online and active. |
What data is sent to the AI
AI Insights analyses the same data you would get by clicking Export on the report. In other words, the CSV view of the report, as currently filtered, is what the AI sees. Nothing else from your Morbit Studio tenant is included.
This means:
- The AI only sees the report you are looking at, for the date range and filters you have chosen.
- The data is operational metadata: call records, device names and status, peripheral types, booking and occupancy figures, sensor readings and quality scores.
- The AI does not receive meeting content. Morbit Studio does not collect call audio, video, chat or transcripts, so none of that can be included.
- The AI does not receive your Morbit Studio credentials, your Microsoft 365 or Zoom credentials, or any configuration data outside the report.
Security and privacy
We designed AI Insights so that using it does not change the security posture of your Morbit Studio data. The key controls are:
Private, in-region AI processing
- AI Insights runs on Amazon Bedrock, a managed AI service within Amazon Web Services (AWS). AWS is already Morbit's hosting provider and is listed as a sub-processor in our Data Processing Agreement. No new third party is introduced.
- The AI service is accessed privately from within the Morbit Studio platform. It is not publicly accessible and there is no public endpoint that could be reached from outside the platform.
- Processing takes place in the same AWS region as the rest of Morbit Studio, within the European Economic Area (Republic of Ireland). Your data does not leave that region to be analysed.
- All data is encrypted in transit between Morbit Studio and the AI service.
No retention, no training
- Each AI Insight is generated on the fly. The report data is sent to the AI, the analysis is produced and returned to you, and the data is then discarded by the AI service. It is not stored by the AI service after the report has been produced.
- Your data is not shared with the AI model's provider. Amazon Bedrock does not pass customer inputs or outputs to the companies that build the underlying models.
- Your data is never used to train or improve any AI model, whether Morbit's, Amazon's or the model provider's.
- Morbit does not build up a history of AI prompts or responses. If you want to keep an analysis, copy it from the page.
Same access controls as the report
- AI Insights can only be run by a user who already has permission to view the underlying report. It never exposes data the user could not already see or export.
- The analysis is scoped to the organisation, site or group selected in the report, exactly as the report itself is.
Contractual commitments
These commitments are written into our Data Processing Agreement, which states that personal data processed by the AI Insights feature is not shared with the AI model's provider, is not used to train or improve any AI model, and is not retained by the AI service after the relevant report has been produced.
How we keep the output consistent
Large language models can produce very different output depending on how they are asked. To make AI Insights predictable and useful across the whole product, every request is built the same way:
- A fixed role.The model is told who it is writing as and for whom. Depending on the report, it is given one of two roles:
- “You are a senior workplace technology and device estate analytics consultant writing for business, operations, and support stakeholders” (used for device, registration, peripheral and call reports), or
- “You are a senior workplace technology and meeting room analytics consultant writing for business and operations stakeholders” (used for room, booking, desk and environmental reports).
- Report-specific guidance. Each report includes its own supporting notes that explain what the columns mean, which thresholds matter (for example what counts as a poor-quality call or an uncomfortable room) and what kinds of recommendation are appropriate. This keeps the analysis relevant to the report rather than generic.
- A fixed output structure. The model is asked to return the summary, risks and concerns, and recommended actions in the same order every time, with owners and timeframes attached to each action.
- The report data. The filtered CSV data for the report, as described above.
Because the role, guidance and structure are fixed, two people running AI Insights on the same report will get analyses that cover the same ground, even though the wording will not be identical.
Getting the best results
- Choose a meaningful date range. A month or a quarter usually gives enough data to show patterns. Very short ranges can produce recommendations based on too few events.
- Filter before you run it. If you want actions for a specific site or group, filter the report to that site or group first. Estate-wide reports give estate-wide recommendations.
- Re-run after changes. Because each analysis is generated fresh, you can change the filters and run again to compare sites, periods or device types.
- Treat it as a starting point. The AI can only see the report data. It does not know about planned works, known faults, building closures or contractual arrangements unless they are reflected in the data. Validate recommendations before assigning them.
- Copy what you need. Analyses are not stored. Paste the output into your ticketing system, review pack or email if you want to keep it.
Frequently asked questions
Which AI model is used?
AI Insights uses Anthropic's Claude models, delivered through Amazon Bedrock. Because the model is hosted by AWS within the Morbit platform, Anthropic never receives your data.
Is my data used to train the AI?
No. Neither Morbit, AWS nor the model provider uses your data to train or improve any model.
Is my data stored by the AI?
No. Reports are analysed on the fly and the data is discarded by the AI service once the analysis has been returned. Morbit does not keep a history of AI prompts or responses.
Does my data leave the EU?
No. AI processing takes place in the same AWS region as the rest of Morbit Studio, in the Republic of Ireland.
Can the AI see meeting content, recordings or chat?
No. Morbit Studio does not collect that content, so it cannot be included. The AI sees only the operational metadata shown in the report.
Does AI Insights see personal data?
Some reports contain personal data such as user names, organiser names or caller identities, in the same way the report and its CSV export do. This is processed under the same Data Processing Agreement as the rest of Morbit Studio, with the additional commitments described above. Where a report offers anonymisation, the anonymised data is what is analysed.
Why does the output differ slightly each time I run it?
Language models generate text rather than retrieving a stored answer, so the wording varies between runs. The structure, role and guidance are fixed, so the substance of the analysis stays consistent.
Can I turn AI Insights off?
Yes. AI Insights is an optional feature. Contact Morbit Support if you would like it disabled for your organisation.
Who can run AI Insights?
Any user with permission to view the report. AI Insights never exposes data the user could not already see or export.
If you have any further questions about AI Insights, security or data processing, please contact Morbit Support.
Was this article helpful?
That’s Great!
Thank you for your feedback
Sorry! We couldn't be helpful
Thank you for your feedback
Feedback sent
We appreciate your effort and will try to fix the article