Verified today
- Individual technical retention rules already exist, but do not yet form an approved overall schedule.
- A general public deletion period would therefore be misleading today.
Security & privacy
VerwaltKlar documents which data is processed for a maintenance case, which services are involved, and where people review or approve. Locations, retention and providers are stated service by service rather than hidden behind a blanket promise.
Public overview
Clear roles. Bounded data flows. Human approvals.
This page describes verifiable product boundaries. It is neither a certification nor a complete disclosure of internal security mechanisms.
Review status: 2 August 2026
At a glance
Access and sharing follow role and case context. AI prepares information; commissioning and required approvals remain with responsible people.
VerwaltKlar prepares the case and suggestions. The property manager selects and commissions the trade business.
Tenant, role and case context limit which data property managers, tenants, owners and trade businesses can see.
We state tasks, data types and locations service by service and label planned building blocks as planned.
Bounded data flow
The public view shows responsibility and data hand-offs, not a reconstructable internal topology.
Telegram is the currently verified reporting channel. Text, optional photos and availability details are connected with the relevant tenancy and property context.
Required report text and bounded case context support the dialogue and structuring. Photos remain case evidence in the current workflow.
The property manager sees the structured details, can edit them and reviews recommendations and required approvals.
A selected trade business receives only the case data intended for acceptance and delivery. A person commissions the work.
AI and model use
For the guided dialogue, required report text and bounded case context are processed with Gemini on Vertex AI. The issue description and relevant lease clause are used to assemble liability information; extracted text is processed for document reading. Ticket photos are currently case evidence and are not described as AI image analysis. Google states that it does not use customer data to train or fine-tune models without prior permission or instruction.
This describes the current division of tasks. It is not a promise that AI output is complete or professionally correct.
Google's information on training and data retentionServices & regions
The core product services for runtime, case data and customer files are currently configured in Belgium; Document AI uses the EU configuration and Cloud Translation uses its EU endpoint configuration. Firebase states that Authentication is processed in the United States, Cloud Logging storage is currently global, and reporting channels and other providers follow their own terms. We therefore make no blanket Europe promise.
Retention & data exit
Before real pilot data is processed, retention, export and subsequent deletion are defined for each data class and service. We publish no general deletion period until that procedure has passed technical and legal review.
Procurement readiness
The public overview deliberately stays limited to commitments, responsibility boundaries and service-specific facts.
For a specific pilot, we jointly map data types, involved services and evidence—through the public overview or, where necessary, in a protected exchange.