Example scenario — not a named customer
A property company asks its own archive
A property company with a mixed residential and commercial portfolio
A property company makes its inspection reports, drawings and leases searchable with an AI server in its own building, with every answer linked back to the source document.
- Business AI
- server tier
- 2× RTX 5090
- GPUs in the building
- 0
- tenant records sent to public AI
The problem
Inspection reports, manuals, drawings and leases were spread across shared folders built up over many years. Answering a question about a building meant searching folders or asking whoever had been there longest, and tenant replies varied depending on who wrote them.
Constraints
Leases and tenant correspondence contain personal data that cannot go to a public AI service. Access has to follow each manager's property portfolio, and the solution has to work with the existing file shares rather than a new document system.
What we deployed
A Business AI server installed on site, connected read-only to the property archive. It provides document search with links to the source, lease summaries, first drafts of tenant replies, and fault reports turned into structured work requests. Access follows the company's existing roles, so each manager only sees their own portfolio.
Outcome
Property managers ask the archive directly — which inspection remarks are still open, what a lease says about maintenance — and open the source document to check the answer. Tenant replies and work orders start from a draft instead of a blank page, and are reviewed before they go out. No tenant data leaves the building.
Configuration
Business AI tier (2× RTX 5090, 128 GB RAM), rented with a maintenance contract