AI analytics
Ask questions about aggregate operational data
Milo — the Graicx AI analytics chat — is designed for scoped, read-only analysis of operational trends, totals, comparisons, and charts.
What AI Analytics can help with
- Work order counts, trends, priorities, statuses, sources, and work types.
- Service request volumes, trends, priorities, statuses, and sources.
- SLA breach analysis by safe aggregate dimensions such as site, client, priority, status, month, or week.
- Asset linkage rates, including linked and unlinked operational records.
- Labour capture rates and hours captured by safe aggregate dimensions.
- PPM analytics, including status, client, site, month, and week breakdowns.
- Compliance and action analytics by status, action type, entity type, priority, site, client, or month.
- Client and site filtered analysis after the assistant resolves the relevant client or site.
- Period comparisons and chart generation from aggregate result sets.
Example questions
- How many work orders did we have last month?
- Show Q1 2026 work orders by priority.
- Compare April vs May work orders by status.
- Show SLA breaches by site.
- Show Shell work orders last month.
- Create a chart.
How results are produced
The assistant calls controlled analytics tools that return aggregate rows, such as counts by status, site, client, priority, month, or week. It is intended for reporting and exploration, not for record creation, updates, or unrestricted database access.
Limits
- No person-level analytics or individual attribution.
- No technician, requester, assignee, employee, or user names.
- No emails, phone numbers, descriptions, comments, notes, or free-text work order content.
- No write actions, workflow changes, or record updates.
- No unrestricted database access or raw operational record browsing.
Aggregate analytics only
AI Analytics is designed around safe aggregate dimensions such as site, client, status, priority, work type, source, month, and week. If a request asks for person-level data, the assistant should redirect to an aggregate view instead.