Sheet 03 — Case Study
Property operations with AI
In active developmentTranquilOps
Property operations for small rental portfolios: one action queue for rent, notices and repairs, plus an AI agent that triages repair requests within rules it cannot break.
Today the app keeps its data in the browser; accounts are coming next.
The problem
Small landlords run rent, legal notices and repairs out of texts, spreadsheets and memory, so overdue rent slips past a notice date and an emergency can sit in the same inbox as a squeaky door.
Who it is for: Operators who manage a handful of rental properties themselves.
What we built
- A daily action queue that puts emergencies first, then legal escalations, approvals and overdue work
- Tenant profiles with balance, payments, messages, the enforcement timeline and an audit trail
- A repair triage agent that sets urgency, picks the vendor, and either dispatches small jobs or asks the operator to approve
Engineering quality
- The AI's rules (a spending ceiling, emergencies always go to a person, no discussion of rent) are enforced in code before and after the model, not only in its prompt
- Each rule has a test that breaks it on purpose and must fail the suite
- An evaluation suite that fails on any missed emergency or broken rule
- Balances computed in exact cents and unit tested
- Automated axe accessibility tests to WCAG 2.1 AA on every page and state, in both themes
Tech stack
- Next.js
- React
- TypeScript
- Tailwind CSS
- Python
- Claude Agent SDK
- Vercel
Every project is scoped to your goals.
Book a call for a tailored estimate.