Ground replies in current listing data
Rent, availability, and property details are retrieved from the business system at the moment of the conversation. The model is not treated as a property database.
AI and automation / 2026
A high-intent property enquiry can arrive after the office closes, sit unanswered until the next morning, and lose momentum while the prospect continues messaging other providers.
The assistant does not invent availability, rent, or property facts from model memory. It reads the listing system, asks two bounded qualification questions, offers a valid calendar slot, writes the lead and booking to operational systems, and hands anything involving tenancy approval, rent negotiation, or screening back to a person. The UK and US variants show the same control pattern with different language and regional conventions.
Walkthrough
Fictional lettings agency. An evening enquiry is answered from listing data, qualified with two questions, booked into a valid viewing slot and written to the CRM and calendar.
My part
Decisions
Rent, availability, and property details are retrieved from the business system at the moment of the conversation. The model is not treated as a property database.
The assistant asks only the questions needed to prepare the team and schedule the next step. More invasive or policy-sensitive screening remains outside the automated flow.
The workflow may place a valid viewing or tour on the calendar, but it cannot approve a tenancy or lease, negotiate rent, or decide whether someone passes screening.
The goal is not a clever chat transcript. The useful outcome is a lead record, a calendar state, an audit trail, and a clear handoff that the team can open the next morning.
Screens
SYSTEM ARCHITECTURE
Website message, form or missed call
Bounded conversation and two qualification questions
Availability, rent and property facts
Valid facts and allowed actions
Wording and extraction of bounded answers
Valid viewing or tour slot
Lead record and booking state
Handoff, tenancy and screening decisions
System components
More work
Gazelle's KitchenA WhatsApp business gatekeeper that qualifies genuine orders, enforces business and safety rules, contains off-topic or abusive conversations, and only surfaces the small set of cases that genuinely need the owner.PilotSellavie AIAn independent project: multi-tenant conversational-commerce software that connects social messaging with business knowledge, guarded AI, human handoff, orders, billing, analytics and operations.MaintainedAI Automation Control PlaneA runnable multi-tenant control plane for AI-assisted business workflows, combining FastAPI, Next.js, PostgreSQL, Redis, n8n, provider routing, human approval, idempotency, replay, cancellation, audit traces, and deterministic local demonstrations.Maintained