Property management agencies lose revenue and trust in the gaps between systems: a slow reply to a rental enquiry, a renewal that nobody chased, a maintenance request that sat in an inbox. We built a lifecycle engine that closes those gaps inside GoHighLevel, and we built it before we had a client to build it for.
The problem
A scan of 140 Perth and WA property management agencies' public websites found the leak was structural, not occasional: 61 agencies showed a clear painkiller signal (slow enquiry response, no visible review flow, thin renewal process) and 47 were ready for direct outreach. The pattern repeats across seven lifecycle moments: rental enquiries answered late, BDM follow-up gone cold, owner communication that only happens when something breaks, maintenance requests chased by hand, renewals and rent reviews missed, and reviews or referrals never asked for. No agency in the scan had one dashboard showing a principal where the leak actually was.
What we built
A GoHighLevel-native sub-account, cloned per agency from a single master snapshot rather than rebuilt each time. The data model centres on one custom Property object with 10 fields, linked to contacts, feeding five native pipelines: Rent Roll Growth, Rental Enquiry, Renewal Radar, Maintenance Follow-up, and Referral/Review. On top of that sit 22 workflows (21 published, 1 in draft), 4 intake forms, 7 email templates, 2 calendars and 8 dashboard widgets giving the principal one view of where enquiries, renewals and maintenance requests actually stand.
AI sits inside this build, not in charge of it. Voice AI answers inbound calls, Conversation AI triages SMS, email and webchat, Reviews AI and an AI data-extraction step draft the busywork. None of it is left to run unsupervised: arrears notices, legal notices, rent-increase comms and screening decisions sit on a hard-coded never-automate list. Those message types always stop for a human to approve before anything sends, enforced by the workflow structure itself, not by a rule someone has to remember to enforce.
Under the hood
The build stays out of the agency's system of record on purpose. It never becomes the ledger, the lease store or the trust-accounting system: it mirrors status and deep-links back, by design, so an agency's existing property management software stays the source of truth. A CSV sync script upserts on a stable external ID and is proven idempotent, tested to created-then-updated with no duplicate records. A 231-line validation script runs after every snapshot import and checks the build's shape before go-live. Consent fields (consent type, consent source, voice-consent expiry) are first-class data on every contact, gating outbound messaging rather than sitting in a policy document, alongside a documented Australian regulatory compliance pass.
What changed
The engine passed its own validator at 93 checks with 0 failures before a single outreach email went out. That is the deliberate order of operations here: build the system once, properly, prove it against real market data, then sell a working thing instead of a proposal. No client has signed yet, and we are saying that plainly rather than dressing up a pre-sales build as a delivered engagement. What exists today is a productised system an agency can turn on, with the judgement calls that matter, arrears, legal, rent increases, still landing on a person's desk before they go out the door.
