LatchFlowAI Operations for LettingsA simple worksheet for lettings teams deciding whether AI should touch inboxes, calls, repairs, compliance chasing, or landlord updates. It helps you see where automation will help and where it's safe to start.
Score each area from 0 to 4. A strong score means the workflow has volume, a clear owner, a clear source of truth, safe approvals, and useful records.
Shared inboxes, calls, WhatsApp, and forms are handled separately with no single owner.
Every inbound item enters one supervised queue with a response SLA and escalation route.
Daily inbound volume, backlog age, missed-call recovery, first-response time.
Staff copy context between the PMS, inbox, spreadsheets, and legacy tools by memory.
One source of truth is named for each workflow, with read/write boundaries documented.
PMS fields, tenancy state, landlord rules, branch ownership, record IDs.
Repair tickets exist, but landlord approvals, contractor chasing, and tenant updates drift outside the ticket.
The full chronology from tenant issue to contractor handoff and landlord update is inspectable.
Issue category, urgency, approval threshold, contractor route, update cadence.
Certificates, renewals, Renters Rights evidence, and missing information are chased ad hoc.
Deadline-sensitive work has owners, proof, reminders, and exception queues before risk becomes urgent.
Due dates, delivery proof, tenancy status, owner approval, missing records.
Payment, invoice, and landlord-update context is visible only to finance or buried in side conversations.
Finance context informs operations without giving AI autonomous payment or accounting authority.
Ledger references, arrears exceptions, invoice state, owner update rules.
Prompts, auto-send rules, confidence thresholds, and overrides are not reviewed after launch.
Human approval gates, override reasons, prompt/rule changes, and rollback paths are logged monthly.
Confidence threshold, override logs, prompt versions, rollback plan, audit export.
By the end of the audit, you'll know whether to begin with inbox triage, maintenance, compliance chasing, or to tidy up your records first.
Current workflow map and system authority boundary
Baseline estimate for inbox, maintenance, compliance, and approval drag
First workflow recommendation with package fit
Approval gates, rollback path, and data access assumptions
Connected-system dependency list with readiness labels
30-day rollout plan and first KPI review template
A low score isn't a failure. It just means it's worth tidying up your process first, so automation lands on solid ground.
Standardise ownership and source records before live AI touches the queue.
Start with one supervised workflow, human approval, and weekly tuning.
Good fit for Core or Pro if queue volume supports the retainer.
Use your score to choose the first workflow, the right package, and the approvals and systems involved before any build starts.
