Worked example
Situation
A small management team receives tenant messages by email and messengers. An administrator works out the property, rewrites the request and looks for a contractor. A repeated message can become a duplicate, and an urgent one can get lost among everyday questions.
What changes
Messages land in one queue. AI suggests the property, category, a short summary and the missing details; rules route the request to the right person. The tenant sees the status.
Where people stay in control
Potentially urgent messages go straight to a person. The system does not promise to recognise every emergency; a clear emergency channel remains.
What changes for the team
Every request shows who is responsible and what has been done, so nobody has to keep the open list in their head.
How the example is calculated
- Before
- 80 h/month
- After
- 35 h/month
Preliminary estimate · month
600 requests × 8 minutes = 80 h; 600 × 3 minutes + 5 h of oversight = 35 h
Assumptions
600 requests a month; 8 minutes of administration now; 3 minutes after; 5 hours of oversight. Repair work itself is not included.
What to check in a pilot
Missed requests, duplicates, correct assignment, time to first human reply. Test urgent requests separately.