Flatt: a property management super-app with AI rent insights

Flatt is a property management platform that connects owners, tenants, and contractors in one app. PHII Labs built the MVP in 10 weeks and shipped the AI layer that handles tenant communications and rent pricing insights.

Context

Property management in the UAE is fragmented. Owners use WhatsApp to talk to tenants. Tenants call property managers for maintenance. Contractors get job details by SMS. Nothing is connected. When a tenant reports a leak, the owner finds out three days later, the contractor gets incomplete instructions, and the property manager is the bottleneck copying messages between everyone.

Flatt was founded to collapse that chain into one app. The team had domain knowledge and a network of property managers willing to pilot, but no engineering team to build the product.

Problem

Three user types with different needs had to share one system:

1. Owners want financial visibility: rent collected, arrears, maintenance costs, occupancy.
2. Tenants want fast resolution: report a maintenance issue, pay rent, ask a question about the lease.
3. Contractors want clear job instructions: what to fix, where, when, and what was agreed.

The communication load is the killer. Tenants message constantly — "when is rent due?", "the AC is leaking", "can I renew for 6 months?". A property manager handling 50 units drowns in this. The product needed to absorb the repetitive communication without losing the human touch on issues that matter.

Approach

We built the app on FlutterFlow with Firebase for state and auth, and Stripe for rent collection. The choice of FlutterFlow was driven by speed: the MVP needed to be in users' hands in under three months to validate the model with the pilot property managers.

The AI layer has two components:

• Rent insights: an OpenAI-powered analysis engine that looks at the property's rent history, market comparables, and lease terms, then surfaces a recommendation — "this unit is priced 8% below market, consider increasing on renewal." Owners see this in their dashboard. It is advisory, not automatic.
• Tenant chat: an AI agent that handles tenant messages. It answers questions about rent due dates, lease terms, and building amenities. It creates maintenance tickets and routes them to contractors. It escalates to the property manager for anything that needs human judgment — disputes, lease negotiations, complaints about other tenants.

The chat agent is built on TypeScript, calling OpenAI with a structured output contract. Every message is classified: answerable, ticket-able, or escalate. The system executes the classification, not the model.

What was built

• Owner dashboard: rent collection status, arrears, maintenance cost tracking, occupancy, and AI-generated rent insights per unit.
• Tenant app: rent payment via Stripe, maintenance request submission with photo upload, and a chat interface that routes to the AI agent first.
• Contractor portal: job assignments with full context — issue description, photos, location, agreed price — pulled from the tenant's maintenance request.
• AI rent insights engine: periodic analysis of portfolio pricing against market data, surfaced as actionable recommendations.
• AI tenant chat: handles routine queries, creates maintenance tickets, and escalates to the property manager with a structured context packet.

Results

• MVP launched in 10 weeks. FlutterFlow plus Firebase plus Stripe gave enough velocity to ship a working product to pilot property managers within the timeline.
• ~120 monthly active users. Owners, tenants, and contractors across the pilot portfolio, using the app for rent payment, maintenance, and communication.
• 60% of tenant chats handled by AI. The chat agent resolved routine queries and maintenance ticket creation without human involvement. Property managers spent their time on the 40% that needed judgment.

Takeaway

A property management super-app lives or dies on the communication layer. The features — rent payment, maintenance tracking, dashboards — are table stakes. What makes the product worth using is that tenants get instant answers and property managers stop being message routers. The AI does not replace the manager; it removes the 60% of messages that never needed a human in the first place.