On this page

An AIPranata concierge is an AI assistant, not a person. It replies and can take actions for the business without a person reviewing each message, and it can be wrong. You can always ask for a person, and businesses decide what the concierge may do on its own.

What a concierge is

A concierge is an AI customer-service agent that a business uses to talk with its customers. It is built on AI models from third-party providers and connected to the business’s systems, so it can look up orders or bookings and take actions the business allows.

Its replies are generated by AI. Unless the business requires approval for a step, a person does not review each reply or action before it happens.

If you are talking to a concierge

You are chatting with an AI assistant acting for the business you contacted, not with AIPranata as a company.

  • You can ask to speak to a person at any time.
  • Check important details, such as refund amounts, delivery dates, or booking times, in the confirmation the business sends you.
  • Do not share passwords, full card numbers, or other secrets in the chat.
  • The concierge may use what you have told the business before to personalise its replies. You can ask the business to correct or delete this information.

What a concierge can get wrong

Concierges work from patterns learned by AI models and the information the business provides. They can sound confident and still be wrong.

  • They can give outdated or incorrect information if the business’s policies or knowledge base are out of date.
  • They can misunderstand a request, or match a customer to the wrong order or booking.
  • They can misread how a customer feels and respond in an unsuitable tone.
  • Remembered preferences can be outdated or wrong.
  • They can make a mistake when taking an action, even within the business’s rules.

How businesses keep people in control

Businesses decide which actions a concierge may take on its own, set limits such as maximum refund amounts, and require approval for anything above them. We recommend handoff rules for upset customers, complaints, requests outside policy, and any situation involving safety, health, or legal matters.

Businesses should review transcripts regularly, keep their knowledge base current, and make sure handoffs reach a person promptly. A concierge should not be the only route to help.

Personalisation and fairness

Personalisation should make help faster and more relevant, not less fair. Concierges may adjust replies using a customer’s history, preferences, and membership tier as the business defines them. They should not treat customers differently because of protected characteristics.

If you think a concierge treated you unfairly, contact the business, and if needed, contact us.

Quality and change over time

A conversation handled under a paid package can still contain mistakes. Your package fee covers access to the platform, not a guarantee that every reply is correct.

AI models, integrations, and concierge capabilities can change. Businesses should recheck important flows after changes to their policies, systems, or products.

If something goes wrong

If a concierge gives harmful or incorrect information or takes an unexpected action, businesses should pause the affected flow and report it to us with the conversation reference and approximate time. Customers should contact the business, which can correct the outcome.

We can investigate concierge behaviour and related charges, but cannot promise to reproduce or reverse every reply, especially messages already delivered.

Questions or concerns?

For questions about these policies, billing, concierge behaviour, or personal data, email us. If you are a customer of a business that uses AIPranata, you can also contact that business directly. Do not send passwords, card numbers, or API keys.

AIPranata

support@aigateway.id