WhatsApp Tenant Support Agent
Adviser involvement in standard queries reduced by 70%; average response time improved from hours to under five minutes.
The challenge
Post-letting tenant support was consuming 130 hours a month in manual WhatsApp responses with inconsistent quality and no triage — a function that could be largely systematised without reducing the quality of complex case handling.
The support inbox had three distinct categories of incoming message, each requiring a different response. The first — routine administrative queries (tenancy end dates, renewal processes, utility setup, council tax exemptions) — had standard answers that didn't vary meaningfully by situation and were being written from scratch each time. The second — guidance on common issues (landlord not responding to repair requests, deposit deduction disputes, neighbour complaints) — required the application of standard UK tenant rights frameworks in a culturally sensitive way, with Turkish-language explanation for clients whose English was limited. The third — genuinely complex or legally sensitive situations — required adviser judgment and sometimes third-party involvement. All three categories were being handled with the same process: a human reading the message, composing a response, and sending it manually. The result was that advisers were spending significant time on the first two categories — work that didn't require their expertise — and had insufficient time for the third, where their expertise actually mattered.
The system
Decision system built
We designed a WhatsApp AI support agent that operates as a first-response layer across the firm's tenant subscriber base. The agent handles category-one queries (administrative and informational) fully, providing accurate, culturally calibrated responses in English and Turkish without adviser involvement. For category-two queries (common issues requiring standard guidance), the agent provides an initial response with applicable UK tenant rights context and notifies the relevant adviser, who reviews the response before it is sent and can amend if needed. Category-three queries (complex, legally sensitive, or emotionally escalated situations) are immediately flagged and routed to an adviser with full conversation context. All interactions are logged, enabling the firm to monitor query type distribution, response quality, and adviser intervention patterns over time.
System components
First-response triage layer: incoming message classification into administrative, standard-guidance, and complex/escalation categories
Administrative query resolution engine: pre-approved responses for standard query types, delivered in the user's preferred language (English or Turkish), personalised with relevant tenancy details from CRM
Standard guidance response drafting: AI-drafted responses for common issue types, routed to adviser for review and approval before sending — not sent autonomously
Escalation routing with context handoff: complex queries flagged immediately to the responsible adviser, with full conversation thread and query classification to enable informed response without re-reading history
Interaction logging and query analytics: full audit trail of all conversations, with query type distribution reporting enabling ongoing improvement of the resolution layer
How we worked
Engagement scope
Support inbox audit and query type classification, AI agent design and response library development, escalation protocol definition, CRM integration for tenancy data personalisation, and performance monitoring setup.
Timeline
Five-week build with a supervised operation period — all agent responses reviewed by an adviser before sending — before autonomous resolution for category-one queries was enabled.
Operating model
The supervised operation period was used to validate response quality and calibrate the classification boundary between categories one and two. The autonomous resolution threshold was raised only when the error rate on category-one responses had been below 2% for two consecutive weeks.
Outcomes
Business impact & measurable results
Adviser involvement in standard queries reduced by 70%; average response time improved from hours to under five minutes.
Average first-response time for routine queries reduced from 4–6 hours to under 5 minutes — the primary service quality complaint from tenants was response speed, and this was addressed without adding headcount
Adviser involvement in support queries reduced by over 70% for administrative and standard-guidance categories — freeing approximately 90 hours per month of adviser capacity previously consumed by routine message handling
Out-of-hours coverage achieved: the agent responds to routine queries at any hour, eliminating the gap between a tenant sending a message at 10pm and receiving a response the following morning
Response quality consistency improved: standard guidance responses are now drawn from the same knowledge base for every subscriber, rather than varying by which adviser happened to respond
Escalation quality improved: complex cases now arrive at the responsible adviser with full context and classification, rather than as unread messages in a shared inbox — reducing the adviser time needed to orient before responding
Governance
Trust, collaboration & governance
Category-two responses (standard guidance) are never sent without adviser review during the first 90 days — autonomy in that category was earned through demonstrated quality, not assumed from the start
Turkish-language responses were validated by a native speaker before being added to the resolution library — accuracy in the tenant's first language is non-negotiable when the subject matter involves their legal rights
Every interaction is logged and available for adviser review — the agent operates transparently, not as a black box processing tenant communications without oversight
Reframe
An agent answering in five minutes at 10pm is not a cost measure — it is a retention mechanism.
Across every engagement, the goal is the same: engineer a system that makes better decisions — faster, more consistently, and at scale — than the process it replaces.
Next steps
Related services
Start a discovery
Most engagements begin with a conversation about context.
We do not send a proposal before we understand the problem. Start by telling us about your decision context — we will identify the highest-leverage intervention areas before any scope is agreed.