Case studies / Student and shared housing operations
Running a 61-bed residence from WhatsApp: rent, KYC and meter readings without an app nobody installs
A residence with 31 rooms and 61 beds runs tenancies, KYC, agreements, rent and electricity readings through WhatsApp instead of spreadsheets and a separate app. Residents pay by UPI and send a screenshot; the system reads the amount and reference and matches it to the ledger. It is self-hosted, with no per-unit SaaS fee.
A shared-housing operator · India
- 61
- beds managed, 31 rooms
- 70
- data models
- 879
- automated API tests
- 0
- apps residents must install
Source: Counted from the production repository and live deployment, October 2026.
What was the problem?
Shared housing generates a constant stream of small admin: who has paid, whose agreement expires, which room's meter reading is due, which resident still owes KYC. In most residences that lives in a spreadsheet and a phone full of payment screenshots someone retypes at month end.
Property-management apps exist, but residents do not install them. They already use WhatsApp.
What did we build?
WhatsApp as the interface
Residents interact through WhatsApp via Meta's own Cloud API, not a reseller. The system checks the 24-hour messaging window before sending, so an undeliverable message is reported as undeliverable instead of silently vanishing.
AI that removes retyping
Vision reads UPI payment screenshots — amount, reference, timestamp — and matches them to the ledger. The same approach reads electricity-meter photos, and a matcher reconciles bank statements against expected rent.
Records that hold up later
House rules are versioned and hashed, so exactly what a resident agreed to can be proven months later. Tenancies, KYC documents and agreements live in one place.
What changed?
- In live use for one property of 31 rooms and 61 beds.
- Payment screenshots and meter photos are read automatically instead of retyped.
- Built to production standard: 70 data models, 48 API modules, 879 automated API tests and 32 database migrations.
What is not working yet?
It runs one property so far. It has not yet been proven across a portfolio of buildings — which is exactly what we would do next with an operator who has several.
Common questions
Can AI read UPI payment screenshots?
Yes. Here a vision model reads the amount, reference number and timestamp from the screenshot a resident sends, and the system matches it to the expected rent. Anything it cannot match is flagged for a person rather than guessed.
Why WhatsApp instead of a tenant app?
Because residents already use it and will not install another app. Sending through Meta's own API, with the messaging window checked first, also means every message is either delivered or reported as failed.
Could this work for your business?
It probably fits if:
- You run hostels, PGs, co-living or other shared housing
- Your rent reconciliation is screenshots and a spreadsheet
- You pay per unit for property software, or have given up on it
What this used
More case studies
Student accommodation
From invisible to 577 search clicks a quarter — and recommended by ChatGPT in 73% of student questions
38.5× search clicks, 15 → 577 per 90 days
Student accommodation — paid search
The Google Ads dashboard said ₹243 a lead. The real figure was ₹1,100. Here is what we found and fixed.
~40% of recorded conversions were fraud
B2B agricultural export
856 indexed product pages for a B2B exporter — and named by ChatGPT and Gemini to international buyers
6.6× search clicks, 58 → 385 per 90 days