AI by sector · Property management
How can AI help a property management firm without replacing its software?
Monday morning at a property management firm. Forty emails are waiting in the shared mailbox, among them a water leak, two disputed service charge statements, lost keys, three invoices from tradespeople and an incomplete tenancy application. Every one sets off sorting, data entry or a reminder. Each of those forty emails is work a supervised AI agent can prepare for your staff.
In brief
Where does AI fit in a property management firm?
Djtal, the operational AI specialist for Swiss companies, installs a supervised AI agent that handles a property management firm's administration (tenant requests, tradespeople's invoices, reminders, preparing minutes and application files). Quorum, ImmoTop2, GARAIO REM or AbaImmo remains the system of record, and approval points, starting with the Swiss Federal Act on Data Protection (FADP), are set in writing at scoping.
Five use cases
Five workflows, each with the agent's share and your team's decision.
| Workflow | What the agent does | What your teams retain |
|---|---|---|
| Tenant requests (email, transcribed phone calls) | Sorts by urgency and subject, identifies the building and the tenant, prepares the ticket and a draft reply | Approves the reply, decides which repairs to order |
| Invoices from tradespeople and suppliers | Reads the invoice, links it to the building and the mandate, prepares the accounting allocation | Checks the cases that fall outside the agreed framework, approves payment |
| Reminders (rent, service charge statements, missing documents) | Spots the due dates, drafts the reminder letters in the right tone for the right stage | Approves each letter before it goes out, handles delicate situations |
| Minutes of condominium (PPE) owners' meetings and condition reports | Transcribes and structures the draft document from your templates | Reads, corrects and takes responsibility |
| Tenancy applications | Checks that the file is complete, extracts the key documents, prepares a comparative summary | Decides who is awarded the tenancy |
According to MIT (2025), 95% of generative AI pilots deliver no measurable return. The ones that survive run on an everyday workflow. The five above fit that profile: requests arrive daily, the rules are known and the gain shows within weeks.
Your software stays
The agent connects to Quorum, ImmoTop2, GARAIO REM and AbaImmo.
A property management firm has already invested years in its management software. The mandates, the properties, the leases and the service charge statements all live there. Swapping it out for the sake of AI would cost more than the AI returns, so the agent is built around it. It reads the mailbox and the documents, uses the exports and APIs where your vendor provides them, and prepares the work directly on your staff's screens where no interface exists.
GARAIO REM already promotes its own AI, and others will follow. We activate what your licence includes, then a dedicated agent fills the gaps, as with Zia at Zoho and the AI built into ERPs.
The method
The success metric is fixed before step one.
- 1.
Map how the work flows today
We spend half a day with your property managers on the workflows where the work piles up, the data they touch (the FADP first) and the interfaces your software exposes.
- 2.
Start with a single pain point
Sorting requests or entering invoices, a single workflow to begin with. We agree the success metric before configuring anything.
- 3.
Compare before and after
Time per case, the share passed to a person, errors avoided. Those figures, taken from your own buildings, settle the next step.
- 4.
Extend one workflow at a time
A new workflow starts with the same framework (written scope, approval at sensitive steps, action log). At the end of the engagement, our embedded engineers hand the controls to your teams.
Nine agents work inside Djtal today, the first since May 2026, and one of them covers our own administration under the approvals and action log of step 4.
Frequently asked questions
Your questions about AI in property management.
Do we have to replace Quorum, ImmoTop2 or GARAIO REM to use AI?
No. Your management software stays the system of record. Djtal's agent reads the mailbox, documents and exports and writes back through APIs, imports or work prepared for your staff. If your vendor offers built-in AI, as GARAIO REM does, we switch it on first and fill in what is missing.
How does the Swiss FADP apply to tenant files?
The Swiss FADP is the first filter of any project. A tenancy file holds salaries, extracts from the debt enforcement register and sometimes family circumstances. Djtal maps which data the agent touches and where it is processed, then limits the scope to the data the processing needs. Hosting and model choices follow from that, documented in writing.
Which tasks does the agent complete by itself, and which does a person approve?
The agent sorts, prepares, drafts replies and keeps the files up to date. Anything that goes to a tenant, an owner or a tradesperson passes through human approval. Latitude grows in writing, one category at a time, once measured cases support it.
How long before the first measurable result?
Expect an initial figure within weeks. A pilot takes one pain point (often sorting tenant requests or entering tradespeople's invoices), a metric defined before the start and a before-and-after comparison. The next processes follow if it holds.
Why start AI in property management with back-office administration, before a tenant chatbot?
Back-office administration comes first because that is where the volume and the gain sit. Every incoming request sets off sorting, reminders, tickets and data entry, and a conversational front end leaves that load untouched. The measurable gain lies behind the inbox, where the hours go missing. A tenant portal can follow, once the back office is keeping pace.
Next step
Your Monday inbox, in half an hour?
Bring a typical week of incoming requests and we go through it together, to find the workflow where an agent pays off most and to see what the Swiss FADP implies for your data. The call carries no commitment.
Last updated: