Property management · United States
AI assistants for a property manager's staff and tenants
A US property manager's staff get answers from company docs in their team chat, and tenants send requests to an AI agent on WhatsApp.
2
assistants: one for staff, one for tenants
for tenant requests, with escalation to staff
The problem
The team’s procedures lived in Notion: how to handle a move-out, who approves what, which form goes where. New staff asked the same questions over and over. Tenant requests came in all day too, and each one needed someone to pick it up and log it.
What I built
Two AI assistants.
The first sits in the team’s chat app. A staff member types a question and the bot answers from the company’s own Notion pages, which are split into chunks and embedded into a Postgres vector store. A second source holds general company knowledge, and the bot picks the right one for each question. It keeps a short history per person, so follow-up questions make sense. Updating an answer means editing the Notion page and re-running the import.
The second is a WhatsApp agent for tenants, built on Twilio. It takes a tenant’s request, logs a ticket and escalates to staff.
The result
Staff get an answer in the chat they already have open instead of waiting for a colleague. Tenant requests are logged and handed to a person, instead of waiting for someone to notice them.