TThis Privacy & Data Protection Policy explains how we collect, use, store, share, and protect information when we provide our AI voice agent and hotline integration services (the “Service”). It is written to help our business customers—and their end customers—feel confident about how data is handled.

1) Who this policy applies to

This policy applies to:

• Business clients (e.g., SMBs) who subscribe to or pilot our Service (“Clients”).
• End users/callers who contact a Client’s hotline and interact with our AI voice agent (“Callers”).
• Website visitors and people who contact us directly.

Where we process personal data on behalf of a Client (for example, handling calls for the Client), the Client is typically the data controller, and we act as a data processor / service provider under the Client’s instructions.

2) What data we may process

Depending on the Service configuration and the Client’s use case, we may process the following categories of data:

• Call data and metadata: call time, duration, routing outcomes (AI resolved vs. transferred), and basic call logs.
• Voice and transcripts: audio recordings (if enabled) and/or speech-to-text transcripts of conversations with the AI.
• Customer service content: messages, questions, and information provided by Callers during the call.
• Client-provided data: scripts, knowledge base articles, FAQs, product/service policies, and (if integrated) CRM fields necessary to complete tasks.
• Identifiers and contact details: phone numbers or account identifiers only if required for the Client’s workflow (e.g., verifying an appointment or order).
• Technical data: system performance logs for reliability, quality, fraud prevention, and debugging.

We aim to minimize data collection and encourage Clients to avoid capturing sensitive data unless it is essential.

3) Sensitive data and payment data

Our Service is designed to handle routine inquiries and basic troubleshooting. Unless explicitly configured and contractually agreed:

• Do not provide government ID numbers, full payment card details, banking credentials, or highly sensitive personal information over the AI line.
• We do not want to store payment card numbers (full PAN), CVV, or online banking passwords.

If a use case requires handling sensitive categories (e.g., health data, financial data), we will assess the risks, apply additional controls, and document responsibilities in a written agreement.

4) How we use data

We use data only for purposes consistent with delivering and improving the Service:

• Provide the Service: answer calls, route to a human agent, perform permitted tasks (e.g., check order status), and maintain call continuity.
• Quality assurance and reliability: monitor latency, call completion, error rates, and system health.
• Client analytics and reporting: produce aggregated metrics like call volumes, resolution rates, and common inquiry categories.
• Safety and abuse prevention: detect spam, fraud patterns, or misuse of the Service.
• Service improvement (optional): improve prompts, routing logic, and conversation flows.

If we use data to improve models or features, we will do so under one of the following:

• Client-controlled mode (default): data is used only to operate the Client’s instance and for debugging/QA, not to train general models.
• Opt-in improvement mode: any use beyond operating the Client’s instance (e.g., broader model improvement) will be explicitly agreed and documented.
  
  

5) Legal basis and compliance approach

We aim to follow applicable privacy laws and good industry practice, including principles commonly reflected in Hong Kong’s PDPO and other regional regulations:

• Purpose limitation: only use data for stated purposes.
• Data minimization: collect only what is necessary.
• Access control: restrict who can access data.
• Retention limitation: keep data only as long as needed.
• Security safeguards: protect data in transit and at rest.

Clients remain responsible for providing appropriate caller notices and obtaining consents where required for recordings, identity verification, or specific processing.

6) Call recording and caller notice

If call recording or transcription is enabled, we recommend Clients provide a clear notice at the start of the call (for example: “This call may be recorded for quality and training purposes.”). We can support configurations where:

• Recording is disabled by default.
• Recording is enabled only for specific queues or time periods.
• Only transcripts (not audio) are stored, or vice versa.
  
  

7) AI behavior, human handoff, and privacy-by-design

Our voice agent is designed to:

• Handle routine conversations and handoff to a human when requests become complex, sensitive, or require manual verification.
• Avoid requesting unnecessary personal data.
• Use guarded prompts and policies to reduce accidental disclosure or unsafe guidance.

We can also configure redaction rules (e.g., masking partial identifiers) in transcripts and logs when feasible.

8) Data sharing and subprocessors

To provide the Service, we may share limited data with:

• Telephony providers (to route calls and enable transfers).
• Cloud infrastructure providers (to host the Service securely).
• Speech-to-text / text-to-speech / LLM providers (to power the AI conversation), when required by the Client’s configuration.
• Customer support tools (ticketing/CRM integrations) if the Client enables them.

We only engage reputable vendors and aim to have appropriate contractual protections in place (e.g., confidentiality, security commitments). We do not sell personal data.

9) Cross-border data transfers

Because some infrastructure or AI vendors may be located outside Hong Kong, data may be processed in other jurisdictions. Where cross-border transfers occur, we apply safeguards appropriate to the risk, such as:

• contractual commitments with vendors,
• access controls and encryption,
• limiting the categories of data transferred.

10) Data retention

We retain data only as long as necessary for the purposes described above, then delete or anonymize it. Retention periods can be configured per Client, for example:

• Call logs/analytics: retained for a set business period (e.g., 30–180 days).
• Recordings/transcripts (if enabled): retained for a shorter configurable period (e.g., 7–90 days).
• Debug logs: retained briefly for security and reliability.

Clients may request shorter retention, or “no storage” modes where feasible.

11) Security measures

We use reasonable administrative, technical, and organizational measures designed to protect data, such as:

• Encryption in transit (e.g., TLS) and encryption at rest where supported.
• Role-based access control (least privilege) and audited access for internal staff.
• Segregation of Client environments where applicable.
• Monitoring for abnormal activity and operational incidents.
• Secure development practices and periodic reviews of system components.

No system is 100% secure, but we design for strong protection and continuous improvement.

12) Client controls and configuration

Clients can typically control key privacy settings, including:

• Whether recordings are stored.
• Retention durations.
• Which integrations are enabled (CRM, ticketing, analytics exports).
• What data fields the AI is allowed to request or confirm.
• When to force human escalation (e.g., complaints, billing disputes, identity verification).

We encourage Clients to implement internal policies and training for their staff who handle escalations and data access.

13) Individual rights and requests

Where applicable, individuals may have rights to access, correct, delete, or restrict processing of their personal data. Because the Client is usually the data controller for hotline calls, requests should generally be directed to the Client first. We will reasonably assist Clients in responding to verified requests, subject to contractual and legal limits.

14) Children’s data

The Service is not designed to knowingly collect personal data from children. If we learn that children’s personal data has been collected inadvertently, we will take steps to delete it or coordinate with the Client to do so, as appropriate.

15) Incident response

If we detect a data security incident affecting Client data, we will:

• investigate promptly,
• take steps to contain and remediate,
• notify the affected Client without undue delay,
• provide relevant details needed for the Client’s compliance and customer communications.
  
  

16) Updates to this policy

We may update this policy as our Service evolves. We will post the latest version on our website and update the “Last updated” date. Material changes may also be communicated directly to Clients.