PropTech · AI assistants
Domovyk
An AI concierge that answers residents of apartment buildings in Telegram — from the building's real data, around the clock.
- Status
- Live
- Year
- 2026
- What we delivered
- Product strategy · AI assistant & MCP server · Telegram bot · Back office · Security & access control · Infrastructure

- AI tools over live building data
- 12
- AI tools over live building data
- neighbours' personal details reaching the AI
- 0
- neighbours' personal details reaching the AI
- automated tests
- 200+
- automated tests
The challenge
Homeowners' associations and property management companies drown in the same questions: how much do I owe, when will the hot water be back, whose car is in my spot, who do I call about a leak. Managers answer them by phone at all hours, while the answers already sit in their accounting systems.
A dedicated resident app is not the fix — nobody installs one more app. And plugging AI into personal data is risky: a chatbot that can see the whole building can leak a neighbour's phone number or debt to anyone who asks.
The solution
Domovyk meets residents where they already are. They write to a Telegram bot — or ask Claude directly — in plain language, and the assistant answers from live building data: charges and debts, announcements, meters, parking, vehicles, service contacts and the association's knowledge base. Residents can even submit meter readings in a chat message.
Under the hood is an MCP server that gives the AI a set of precise, permission-aware tools instead of raw database access. A resident is identified by their phone number — no passwords, no sign-up — and one person can own property in several buildings. Managers get a back office to run organizations, staff roles, privacy rules, announcements and meter readings, with data imported from the accounting platform they already use.
Security is designed in from the first line: every request passes through a single access-policy layer, organizations are fully isolated, an assistant never gets more rights than the person behind it, and neighbours' personal data is removed before it ever reaches the model.
Key capabilities
- 01
Answers from real data
Debts, announcements, meters, parking, vehicles and contacts come straight from the building's records — the model never guesses.
- 02
Zero-install for residents
Telegram or Claude, signed in by phone number. No app, no password, no onboarding.
- 03
Privacy by architecture
Neighbours' names and phones are redacted in the data layer, not by prompt instructions — the AI simply never receives them.
- 04
Multi-tenant by design
One platform for many HOAs and management companies: isolated organizations, staff roles scoped down to a single building.
- 05
Full audit trail
Every access records who looked at what, when and through which channel — visible to the organization.
- 06
Back office for managers
Announcements, meter readings, contacts, knowledge base, privacy rules and imports from the existing accounting platform.
Screens


The result
Domovyk is live at domovyk.co.ua and answers real residents in Telegram. The platform is backed by more than 200 automated tests.
For a management company, connecting a new building is a data import, not a project: residents just ask in the chat instead of calling the office, and every data access stays on an audit log the organization can review.
Technology
- TypeScript
- Model Context Protocol
- Claude API
- Node.js
- PostgreSQL
- Hono
- React
- grammY
- Podman
- Caddy
Want something like this for your business?
Tell us about your idea or the process you want to automate. We'll propose a solution, a timeline and the first working version.
Next case
Plate Gate
License plate recognition that turns ordinary IP cameras into a smart parking log — every pass on video, every car accounted for.