Skip to content
Log in
Get started

AI Use Policy

Last updated: 31 October 2026

This policy will be reviewed by counsel; we will tell you before any change. It forms part of the Terms of Service.

This policy explains how BOOK IN CHAT PTE. LTD. (UEN 202646792Z) (BookInChat) uses artificial intelligence (AI) in its service for salons, wellness shops and skin and aesthetic clinics (Shops), what Shops must do, and how Customer data is and is not used to improve our AI.

1. Where we use AI

  1. Drafting and sending replies to Customers' WhatsApp messages: answering questions about services, prices, opening hours and location from the information the Shop provides, and helping Customers book, change or cancel appointments.
  2. Understanding messages: transcribing voice notes, reading images Customers send (for example a payment receipt), detecting the language, and recognising requests such as bookings, complaints, emergencies or STOP.
  3. Safety routing: detecting messages that need a person (clinical questions, emergencies, complaints, sensitive topics) and handing them to the Shop's staff.
  4. Staff tools: suggested replies, conversation summaries and search for the Shop's staff.

BookInChat uses one or more of the AI providers on the Sub-processor list: NUS SoC LaaS (National University of Singapore School of Computing), Alibaba Cloud Model Studio (Qwen), Google (Gemini API), OpenAI (OpenAI API) and Anthropic (Claude API). They are reached only through BookInChat's own AI gateway (LiteLLM) and are used for every Shop, including clinics, where they read the health details patients share after agreeing to the clinic's notice. Images from clinic patients are never sent to an AI model. These providers may process data outside Malaysia and Singapore; we rely on contractual terms with each provider and on its published API data terms, and we use only API services whose published terms say the provider does not use our data to train or improve its models without our agreement (NUS SoC LaaS has no published API terms; we use it under our access arrangement with NUS). We give Shops at least 3 hours' notice before adding or replacing an AI provider.

2. Human oversight

  1. The Shop decides how much the AI does. In its settings, the Shop chooses whether the AI may reply automatically and which topics go to staff.
  2. The Shop approves what the AI may say. The AI answers from the menu, prices, policies and FAQ answers the Shop has entered and approved. The Shop must review them before switching on automatic replies and keep them up to date.
  3. Clinics must approve AI answers. For clinics, the AI may only send answers to health or treatment questions that the clinic has approved in advance (approved templates and FAQ answers, sent as approved); other clinical questions go to the clinic's staff. Approvals are recorded with the approver's name and role. Automatically sent text never names prescription products by brand.
  4. Staff can take over at any time, and Customers can ask to speak to a person at any time.
  5. BookInChat monitors quality, investigates reported errors and improves safeguards, but does not review every conversation.

3. Accuracy

AI can make mistakes. It may misunderstand a message, give an outdated price or time, or state something with more confidence than it should. BookInChat designs the service to reduce errors (answering from the Shop's own information, confirming bookings against the live calendar, and handing over uncertain cases), but cannot promise that AI output is accurate, complete or suitable. The Shop is responsible for checking its information and for the replies sent in its name, and should correct errors promptly when it sees them.

4. No medical, legal or financial advice

  1. BookInChat's AI does not give medical advice, diagnose, recommend treatments or medicines, or assess whether a treatment is suitable for a person. It gives general booking and service information and directs clinical questions to the Shop's qualified staff.
  2. The assistant does not show emergency numbers or crisis helplines. When a Customer's message contains urgent words (for example a possible emergency, self-harm or a problem after a treatment), BookInChat alerts the Shop's staff, for every type of Shop, and the assistant offers to alert the Shop's team; if the Customer says yes, the alert is raised as urgent. The AI does not try to assess these messages; the Shop's staff decide what to do.
  3. The AI does not give legal or financial advice.
  4. Clinics remain responsible for regulatory compliance, including healthcare advertising rules, professional conduct rules, consent for treatment and patient records. BookInChat is not a healthcare provider.

5. Telling Customers they are talking to AI

Customers are told that replies may be drafted by AI (see our notice for customers). If a Customer sincerely asks whether they are talking to a person, the reply must be truthful. Shops must not present AI replies as personally written by a named doctor or practitioner.

6. What we do not use AI for

We do not use AI to make decisions about Customers that have legal or similarly significant effects (for example refusing service or setting prices for an individual) without a person at the Shop deciding. We do not use AI alone to suspend or end a Shop's account.

7. Using data to train and improve a Shop's own assistant

  1. What we do. Unless the Shop opts out, BookInChat may use de-identified data derived from a Shop's conversations and bookings to train and improve that same Shop's assistant (for example its prompts and tuning). This includes fine-tuning our own AI models, on our own servers or, in future, in a data centre we operate. A Shop's data is never used to train or improve another Shop's assistant or any shared model. De-identified means we remove or replace names, phone numbers, WhatsApp IDs, addresses, identity and payment numbers and other personal identifiers, generalise dates and rare details, discard images and audio, and keep no key linking the data back to a person. The data is tagged with the Shop it came from. Details: Data Processing and Shared Control Agreement, clause 6 and Schedule 2.
  2. Health and sensitive data are always excluded. All data from clinics, and any conversation that appears to contain health, medical or other sensitive information, is never used for improvement, even de-identified and even if the Shop has not opted out.
  3. Opt-out. A Shop can opt out at any time by emailing support@bookinchat.com, free and with no loss of features. We then stop using that Shop's data and remove the de-identified data tagged to it.
  4. Customers can object. A Customer can object to this use through the Shop or by emailing support@bookinchat.com. The objection takes effect within 5 business days, and that Customer's conversations are no longer used.
  5. No re-identification. We do not attempt to re-identify anyone from de-identified data and we prohibit others from doing so.
  6. Retention. De-identified data tagged to a Shop may be kept after the Shop leaves, unless the Shop opted out.
  7. WhatsApp platform terms. Meta's terms for the WhatsApp Business Platform allow data from the platform to be used to improve an AI model only for the business's exclusive use. Keeping improvement to each Shop's own assistant follows that rule.

8. Governance

BookInChat's owner and Data Protection Officer, Lee Cheng Han, is accountable for AI use. We keep a record of the AI models and versions in use, test changes before release, log safety hand-offs and review incidents. Report AI errors or concerns to support@bookinchat.com.

9. Changes

We will give Shops at least 14 days' notice of material changes to this policy. A Shop that does not agree may cancel before the change takes effect.

AI Use Policy | BookInChat