Case study · MZGRADA
MZGRADA
AI manages apartment buildings
The first Aseler product in production. Building managers decide faster, residents get verified prices, voting is cryptographically transparent.
Problem
Building managers in Serbia deal with an information asymmetry that costs every resident hundreds of euros per year. Supplier quotes arrive on paper, voting happens at meetings no one attends, and the Housing Act has been read by three people in the country. The result: overpriced paint jobs, muddy minutes, endless email threads.
Solution
MZGRADA is a building management platform that introduces three things. First, a market price database: the agent checks every quote against real market prices and flags overpriced bids. Second, cryptographically secured voting with SHA-256 receipts, where every vote has an audit trail. Third, a specialized AI agent for the Housing Act and mandatory maintenance investment fund, answering residents' and managers' questions from actual articles of the law.
Technology
Next.js 14, TypeScript, PostgreSQL, pgvector for the Housing Act semantic database, AWS SES for notifications, WebAuthn for passwordless login. Hosted on AWS Frankfurt, GDPR-compliant.
Results
In the first six months the platform rolled out in pilot buildings in Novi Sad. Managers report that decisions which used to take weeks (collecting quotes, meetings, signatures) now close in 2-3 days. Residents see the full audit trail of every decision.
Want something similar for your business?
If you have a coordination problem where time and transparency cost money, let's talk.