Google Cloud
A secure, scalable operating foundation in the europe-west1 region, provisioned as code.
Technology
Modern AI, applied in a way you can follow.
VerwaltKlar uses Gemini on Vertex AI and other Google Cloud services to support the work that happens by hand today: listening, asking follow-ups, classifying, translating and reading documents. Decisions about cost and commissioning stay with people.
Data stays in Europe.
Privacy, role-based access and human approval are built into the operating model.
Technology foundation
Powerful technology, kept deliberately simple.
Customers should understand what the technology does for them without needing a map of the underlying infrastructure.
Gemini on Vertex AI
Supports conversations, understanding and triage while decisions remain with people.
Where the AI works
Six places where AI replaces real manual work.
Gemini on Vertex AI handles the AI interactions in the product, with Document AI for documents and Cloud Translation for languages. Nothing is there for its own sake — each point replaces a step someone in the office does today.
- 01
The guided conversation
The chat understands what the tenant means, even in broken German or another language, and asks targeted follow-ups until the report is complete.
- 02
Classifying the report
Trade, affected unit and a usable title are derived from text and photos — instead of 'problem in flat', it says what actually needs doing.
- 03
Pre-assessing urgency
The AI proposes a priority and explains it. Safety-relevant cases are additionally escalated by a fixed rule, never on a model's judgement alone.
- 04
Pre-assessing cost liability
The AI compares the report against the lease's liability clauses and proposes whether the case falls to the owner or the tenant — a decision aid for the office, never a ruling.
- 05
Bridging languages
The report is translated automatically into the reader's language. The tenant writes in theirs, the office and the trade read in theirs — and the original stays available at all times.
- 06
Reading documents
Leases and invoices are read through OCR and AI and turned into structured fields — the basis for tenant onboarding and the tax-report export.
The flow
From a message to a record.
- 01
A message in the messenger
The tenant writes as they would anyway — in their own words.
- 02
Guided conversation
The AI asks the missing questions and collects photos and time windows.
- 03
A structured ticket
The conversation becomes a record with category, location and priority.
- 04
Matching
The case is suggested to a suitable business, and the office approves it.
Guardrails
What the AI explicitly does not do.
- It never commissions a tradesperson on its own.
- It approves no costs and makes no budget decisions.
- It never overrides the property manager's assessment.
- It never replaces the fixed escalation rule for safety cases.
Suggestions are visibly labelled as suggestions and can be overridden at any time.
Security and data protection
Why a German property manager can trust us with tenant data.
Tenant data is sensitive. That's why access isn't a matter of trust but of engineering: every access is bound to a role and checked on the server.
- 01
Data stays in Europe
Processing and storage take place exclusively in European data centres — services are pinned to the europe-west1 region.
- 02
Role-based access
Office, owner, tenant and tradesperson each see only what their role requires — a business sees only its own case.
- 03
Checked on the server
Permissions are enforced server-side, not in the interface. Sensitive operations never run straight from the client.
- 04
Stored encrypted
Data and attachments are encrypted in storage and in transit; credentials are managed separately and protected.
- 05
Transparency toward tenants
The chat tells the tenant at the outset that their details are shared with the property manager.
- 06
Privacy by design
The process is aligned with GDPR requirements — purpose limitation, data minimisation and traceable sharing.
What's next
What we're building next.
The structured tickets are the foundation. The next stages build on them — partner property managers help decide the order and the weighting.
- Additional context from building and asset history
- Energy-certificate data per building and unit as a first-class record
- Multi-entity support for managers running several companies
- Predictive maintenance based on recurring cases
Evaluate the product in a real maintenance workflow.
A partnership gives your team direct access to the product and its safeguards.