After-hours answer bot for property managers

The idea

Property managers absorb the same questions on repeat — when is rent late, can I have a dog, who fixes the disposal — and the questions arrive at 9pm because that’s when tenants are home. The answers already exist in the lease and house rules; the PM is a human retrieval system with no off switch. Support bots are established spend, with Chatbase at $150/mo, but that pricing and packaging targets SaaS companies with websites, not a PM juggling 80 units over text message.

The wedge is grounding plus channel plus triage. Tenants text — they will never install a portal app — so the bot lives on SMS and knows who’s texting from the phone number. It answers from the actual lease with a citation, and the piece generic bots can’t do: it classifies maintenance messages by urgency and phone-calls the on-call number for a burst pipe while quietly ticketing the dripping faucet. That triage is the 2am value that makes a PM pay.

Who pays

Independent property managers and small PM firms with 20-200 units, where after-hours calls land on the owner personally. $79/mo for a small portfolio undercuts generic bot platforms and costs less than one hour of answering-service time. Distribution: PM associations and BiggerPockets-style landlord communities, plus integration listings alongside Buildium and AppFolio ecosystems later.

MVP scope

• Per-property document upload with per-unit lease details and pgvector retrieval
• Tenant roster with phone-number identification over one Twilio SMS number
• Cited answers with a low-confidence fallback that tickets instead of guessing
• Three-tier maintenance triage with phone-call escalation for emergencies
• PM dashboard: conversations, tickets, gap report, pinned corrected answers
• Skip for v1: voice answering, rent payment handling, vendor dispatch, multi-language, Buildium/AppFolio sync

Build LeaseLine, an SMS answer bot for property managers grounded in each property's actual documents. Stack: Next.js (App Router), Postgres with pgvector for retrieval, Claude API for answering and triage, Twilio for SMS.

Core flow: 1) PM creates properties and uploads the lease template, house rules, amenity info, and per-unit lease specifics (rent due date, pet clause, parking spot); documents are chunked and embedded per property. 2) PM adds tenants with phone numbers mapped to units, and gives out one Twilio number per portfolio. 3) Tenant texts a question — can I paint the bedroom, when is rent considered late, how do I book the elevator; the bot identifies the tenant by phone number, retrieves from that unit's lease first and the building docs second, and answers with a citation like: per section 8 of your lease. If retrieval confidence is low it says it will check with the manager and opens a ticket instead of guessing. 4) Maintenance messages get triaged by Claude into emergency (gas, flood, no heat in winter — immediately call and SMS the on-call number via Twilio), urgent (ticket plus next-morning digest), or routine (ticket). 5) PM dashboard shows conversations, tickets, unanswered-question gaps to fill with new docs, and an answer-correction flow that pins a manager-approved answer for future reuse.

Pages: properties and documents, tenants, inbox/conversations, tickets board, escalation settings, analytics. Non-obvious details: never answer legal-dispute or eviction questions — hard-route those to the PM; log every answer with its source chunks for liability. Pricing: $79/mo up to 50 units, $149/mo up to 200.