PROPERTY OS
Built at Meta Global AI Developer Hackathon · Oct 26, 2026 · Remote

The problem According to Buildium, what U.S. renters want most is fast, reliable responses to maintenance requests, and according to AppFolio, renters who are satisfied with maintenance are 81% more likely to renew. The rest of the process doesn’t help either: also per AppFolio, three-fourths of renters have faced challenges during move-in, and per Entrata, 40% completed the entire leasing process in person. On payments, the Philadelphia Fed reports that about 20% were unable to pay rent on time or in full, and AppFolio adds that 39% still use traditional payment methods. When a tenant leaves, Zego estimates a cost of $3,872 per turnover, RealPage reports 34.4 vacant days per unit, and Apartment List puts national vacancy at 7.3% in 2026. Alveole’s analysis calculates that, in a 150-unit building, annual losses reach $260,000-$300,000, and Zego reported that retention fell from 60% (2024) to 57% (2026). 2026 renters report +10 What PropertyOS AI does PropertyOS AI is a multimodal, agent-based platform that centralizes property operations for tenants, property managers and owners. Each user sees only what their role allows, and sensitive decisions such as spending, repairs and contract changes always wait for human approval. AI Chat (Meta Muse Spark 1.3): tenants ask about their rent balance and maintenance requests, and report problems in plain language. The agent answers using only authorized tools and the data of the user’s own property. Maintenance reports with images (Meta SAM 3.1): a tenant uploads a photo, for example of a leak, and SAM 3.1 highlights the affected area as visual evidence. Muse Spark combines the photo and the description to assess urgency and recommend the right specialist. The segmentation supports the assessment; it is not a definitive diagnosis. Voice reporting (Meta Muse Voice Transcribe): a tenant speaks the problem, reviews and edits the transcript, and sends it into the same flow. Supervised maintenance workflow: tenant report → AI triage → specialist recommendation → contractor quotations → owner approval. Finance: a rent ledger tracks obligations, and a payment is reconciled only after the provider verifies it. PayPal Sandbox order creation is tested, but the full buyer-approval-to-capture flow is not yet validated. Leasing and Contracts: managers handle availability, tour requests and applications, and AI-extracted contract terms go through human review before any lease record changes. Supervisor: managers can review agent execution traces, tools used, policy constraints, pending approvals and audit logs.