Phone and text, one single entry point.
Phone (SIP / PBX), Web Chat, WhatsApp, website embedding and the admin console are unified, so customers get consistent answers on every channel.
ZONNET AICC · SMART CUSTOMER SERVICE
Let AI hear and handle every call and message first.
Phone and text channels, trilingual service in Cantonese, Mandarin and English, with human takeover at any moment. Built for high-volume service scenarios such as utilities, hotels and property management — delivered as a SaaS plan or on-premise deployment.
Unified text and voice capability across the full customer-service journey.
Hand repetitive enquiries and services to AI, keep work that needs judgement and communication with people — all connected to the same knowledge base and ticketing workflow.
Cantonese, Mandarin and English — all understood.
Trilingual speech recognition and intent understanding identifies enquiries, repair requests, complaints and other service goals, with sentiment and urgency classification.
Answer the question, then get the job done.
Common enquiries are answered instantly from the knowledge base; repair, complaint and booking requests automatically create tickets, notify handlers and track status through to confirmation.
When it should be human, it is human.
Confidence-based automatic hand-over to agents; emergencies, escalated complaints and safety risks go straight to people, with full conversation context in the agent workspace.
ARCHITECTURE
A five-layer architecture, from channel access to data security.
Built around a large language model (LLM), RAG knowledge base, intent recognition and ASR / TTS voice capabilities — connecting customer channels above and enterprise systems below into a complete service loop.
- Channel access layer: phone / PBX, Web Chat, WhatsApp, admin console
- Conversation orchestration layer: session management, channel adaptation, human hand-over, conversation records
- AI capability layer: LLM, intent recognition, RAG knowledge base, ASR / TTS
- Business service layer: enquiry answering, ticket management, complaint handling, reporting
- Integration & security layer: PMS / CRM / telephony integration, audit, encrypted backup
Phased deployment, with each phase delivering value on its own.
Go live without deep system integration first, lowering upfront risk and letting management see the real impact of AI customer service step by step.
Research & PoC
Map channels, high-frequency questions and the system environment; validate AI answer quality on a small scale.
Pilot Run
Connect real customer channels and human hand-over, with the agent workspace and management reports.
Full Launch
Permissions, audit, backup and training in place, with measurable acceptance metrics.
Deep Integration
Integrate PMS / CRM, telephony and business APIs as needed to close the automation loop.
Designed for Hong Kong PDPO and the PCPD AI governance framework.
Privacy and security are not add-ons but core parts of the solution — from data minimisation to audit and backup.
Privacy Compliance
- Data minimisationOnly the data needed to complete the service is collected; identity documents and payment data never enter the LLM context.
- Usage disclosureConversations open with a clear notice that AI customer service is in use, with a human option available at any time.
- Human oversightComplaints, refunds, identity verification and safety incidents must be handled by people.
- Data retentionRetention periods for conversations, recordings and tickets are confirmed with the client, with automatic deletion or anonymisation on expiry.
Security Assurance
- Permissions & auditData visibility is controlled by role, with full audit logs of administrative actions.
- Transport & storageSite-wide encrypted transport, encrypted storage of sensitive fields, regular backups and recovery drills.
- Model securityClient data is never used to train third-party models, with protection against prompt injection.
- Data residencyHong Kong data residency and high-availability (HA / DR) deployment available on request.
Built for high-volume, repetitive scenarios that need fast triage.
Already deployed in high-volume service scenarios such as the power sector, and expanding into hotels, property management and other industries — further scenarios can be assessed against each client's actual workflows.
SMART CUSTOMER SERVICE
FAQs and service enquiries
Room facilities, policies, billing and other high-frequency questions get a consistent, fast first response.
BUSINESS WORKFLOWS
Repairs and ticket triage
AI gathers details, creates tickets, notifies handlers and updates status through to customer confirmation.
HUMAN-AI COLLABORATION
Agent assistance
Customer intent and context carry into human service, cutting repeated communication, with full supervisor oversight.
Tell us where your customer service gets stuck.
Share your call and message types, existing systems and manual workflows — together we can identify the AICC scenarios to deploy first, and choose a SaaS plan or on-premise deployment.