Industry · Housing and property management

Software for a property management company: requests, deadlines and work reports

Residents' requests from every channel in one window. The AI determines the type of work, the system tracks the deadline, the technician reports by voice

What gets in the dispatchers' way

Residents post on the city portal, call the dispatcher, send emails and chatbot messages. Each channel lives in its own window, and the dispatcher copies requests by hand, picks a team and watches the deadlines. Overdue requests on the city portal are visible to the regulator, and the company is held responsible for them.

The second problem is the category. Residents pick a topic however they like: a leak in the basement gets filed under “damage to building elements”, and the request goes to the wrong team with the wrong deadline.

What the system does

  • One dispatch desk for all channels. Connectors pull requests from the portal, email and the chatbot; the dispatcher adds phone requests with one button.
  • The deadline is visible at once. Each request shows how much time is left before the deadline under the regulations.
  • Type of work from the request text. The AI reads the request and determines the type of work from the company's catalogue. If the resident picked the wrong topic, the system recalculates urgency and the deadline.
  • A team in one click. The system picks the responsible team by building and type of work; the dispatcher confirms.
  • A work report from a voice note. The technician takes before and after photos and says what was done. The report is assembled on the company's template: address, works, materials, time.
  • A reply to the resident. Once the report is signed, the resident gets a reply in the channel the request came from. The AI drafts the reply as well.

How we checked the accuracy

Before promising any accuracy, we built a reference set: more than 150 real requests with labels checked by hand. We ran several models on it. The type of work is identified correctly in about 98% of cases, and every emergency request was recognised as an emergency. When confidence is low, the request goes to the dispatcher for manual review.

4 channelsof requests in one window
≈98 %correct type of work on the reference set
30+buildings under management

Residents' personal data does not reach the model: the text is anonymised before sending. The property management company chooses the model.

Who it's for

For property management companies, homeowners' associations and maintenance services that receive requests from more than one channel. We start with a demo on your typical requests and work out how much dispatcher time it frees up.

FAQ

Which request channels can be connected?

The city portal, email and the chatbot are connected through connectors, and the dispatcher adds phone requests with one button. If you have a channel of your own, such as a mobile app or a messenger, we will add one more connector.

How accurately does the AI identify the type of work?

On a reference set of more than 150 real requests labelled by hand, the type of work is identified correctly in about 98% of cases, and every emergency request was recognised as an emergency. When confidence is low, the request goes to the dispatcher for manual review.

Can we use our own catalogue of works and teams?

Yes, the system works from the company's own catalogue: types of work, deadlines under the regulations and teams, which are picked by building and type of work. After each change to the catalogue the accuracy check can be rerun.

Do technicians need to install an app?

No. The technician's app opens in the phone's browser. The technician sees the work order, takes before and after photos and says what was done; no typing in gloves.

What about residents' personal data?

The request text is anonymised before it is sent to the model. The property management company chooses the model: Russian, foreign or local.

How long does the launch take?

First we show a demo on your typical requests and work out how much dispatcher time it frees up. A pilot on real data goes live within a month; we estimate the cost after a call.

Send us a few typical requests. We will show how the system parses them and work out how much time your dispatchers will get back.