AI Services and Automation Terms

AI systems generate responses based on patterns and the information they are given. They can produce responses that are incorrect, incomplete, out of date or misleading, even when trained on accurate source material. Cailub does not guarantee that AI systems will never make mistakes.

You must review business-critical outputs before relying on them, keep appropriate people involved in important decisions, and monitor your assistant regularly. You must not rely on Cailub for professional legal, financial, medical or other regulated advice unless that is part of an appropriately approved service arrangement.

You are responsible for the accuracy of the information your assistant uses, for approving the content it answers from, and for the instructions, restrictions and tone you ask us to configure. Tell us promptly when your source information changes.

Your instructions are one layer of how your assistant works. Cailub’s platform and safety instructions always take priority over them, and over anything in your knowledge sources or said by visitors and callers. You can edit your own business instructions, see their history and restore an earlier version; you cannot edit or remove Cailub’s platform instructions, required AI disclosures or the “Powered by Cailub” attribution.

You can keep the default assistant name (Cailub for chatbots, Laila for the AI receptionist) or choose your own. A custom name only changes what the assistant calls itself in conversation; names that could mislead people may be rejected or held for review.

Each assistant should have clear escalation rules, such as passing complex, sensitive or urgent questions to a person and telling users how to reach your team. We will help you configure these; you are responsible for approving them and staffing the escalation route.

Where an assistant or agent connects to other systems (such as your inbox, calendar or CRM), each permitted action must be clearly configured and authorised by you in writing. Integrations use the least access needed.

AI agents must not carry out high-risk or irreversible actions, such as moving money, deleting important records, entering binding commitments or making consequential decisions about people, without suitable permissions, safeguards and human approval.

Our AI services rely on third-party models, currently from Anthropic. Model behaviour, availability and pricing may change. We may change or update models to maintain quality, security or availability, and we will tell you about material changes. We cannot guarantee that a particular model will remain available.

Every Cailub chatbot tells visitors they are talking to an AI: an "AI assistant" label (you can reword it but not remove it) under your business name, and a note that AI can make mistakes beside the message box. Replies are also marked as AI-generated in a machine-readable way. These notices are part of the product in every market and cannot be removed or hidden by configuration.

A chatbot cannot go live until you add a link to your own privacy notice. Visitors see it in the chat and in the "Leave your details" form.

You must not configure an assistant to claim or suggest that it is a human.

Fair use on chatbot plans

Each chatbot plan includes up to 5,000 new website conversations a month (fair use). We email you at 75% and 90%. At 100%, new conversations pause until the 1st of the next month; conversations already under way can finish. Nothing extra is charged. If you regularly need more, contact us and we will suggest a plan that fits.

Installation service

Installed for you is a one-off installation fee shown before you pay. When Cailub installs both your chatbot and Laila, they are one project and the installation fee is charged once, whichever you buy first.

Add-ons are monthly subscriptions that work with an active Cailub chatbot plan. There is no setup fee and you can cancel at any time in your billing settings; the add-on then stays on until the end of the month you have paid for. If your chatbot subscription ends, your add-ons are set to end at the close of their current paid month, so you are not charged for something that can no longer work.

AI Appointment Booking

Booking works with Google Calendar. You connect your own Google account and choose the calendar; you can disconnect at any time. A booking is confirmed to a visitor only after Google Calendar has accepted it. The assistant itself never books, moves or cancels appointments: visitors do that through the booking form or the link in their confirmation email, and you can cancel from your dashboard.

You are responsible for keeping your hours, appointment types and calendar accurate. We take reasonable steps to prevent double bookings, including re-checking your calendar at the moment of booking, but events added to your calendar outside Cailub at the same moment, or Google being unavailable, can affect this. If Google Calendar cannot be reached, nothing is booked and the visitor is told.

Confirmation, change, cancellation and reminder emails are sent to visitors on your behalf. They are service messages about the appointment, not marketing.

AI Human Handover

When a visitor asks to speak to a person, or the assistant brings in your team, the assistant stops replying until your team hands the chat back or closes it. Messages from your team are shown with the team member’s first name and clearly marked as coming from a person.

Outside the hours you set, visitors are told nobody is available and can leave their details. You must not configure messages that suggest a person is available when nobody is, and team members must not pretend to be the AI assistant, or the assistant a person.

You decide who on your team can read and reply to chats. Each person must use their own account, and you are responsible for what your team writes and for removing people who no longer need access.

Laila is the name of Cailub’s AI assistant. It is an AI system, not a person, and it identifies itself as an AI. A business may give it a different name.

AI Phone & SMS is a monthly subscription for calls answered by an AI voice agent, text messages, or both. Each plan includes a monthly allowance of call minutes and texts. Choosing both voice and SMS adds a separate monthly charge, shown before you pay. Phone numbers are rented separately, at the monthly price shown before you choose one, and appear as their own line on your invoice.

Usage above your allowance

Extra minutes and texts are charged only above your allowance, at the rates shown on the pricing page when your billing month starts, and are added to your next invoice. Texts are counted in segments (a long message or one with special characters can use more than one), and texts to some countries use more than one unit of your allowance; this is shown in your dashboard. We alert you at 75%, 90% and 100% of your allowance. You can choose to stop at your allowance, or to continue with an optional monthly limit on extra charges.

Numbers and countries

Availability of numbers, number types and texting differs by country and is shown before you choose. Some numbers need a business address or verified documents before they can be activated, and some countries require carrier registration before business texting can start; nothing is charged for a number until it is activated. A number is only shown as active after a real test call has reached your agent. You can keep your existing number by forwarding calls to your Cailub number. Moving (porting) a number in or out depends on the country and the carriers involved and is not guaranteed.

Your responsibilities

• The agent always tells callers it is an AI assistant, and announces when calls are recorded. You must not configure it to suggest otherwise.
• You decide whether calls are recorded and must follow the rules on recording calls in your country and your callers’ countries.
• Outgoing AI calls may only be made to people who asked to be called, or to existing customers about their booking or service. Cold calling and marketing calls are not permitted.
• Marketing texts may only be sent to people whose consent you have recorded, or where the law allows it for existing customers. You must honour opt-outs, which Cailub applies automatically, and identify your business in each message.
• Texts and calls are not sent outside reasonable local hours, and the agent does not handle emergencies: it is not a replacement for emergency services.
• Payment requests by text are only available where a payment integration has been set up.

Phone and text delivery depends on telephone carriers outside our control. We do not guarantee that every call will connect or every text will be delivered, and we show delivery failures in your dashboard.

A new chatbot is called "Cailub" and shows "Powered by Cailub" at the bottom of the chat window. You can rename it, choose your own colour and switch off "Powered by Cailub" in your chatbot settings at any time.

While "Powered by Cailub" is on, the assistant may say that the chat is provided by Cailub if a visitor asks. When it is off, the assistant does not mention who built it.

When you buy a chatbot plan, we create your account area and add the website address you gave at checkout. You then confirm that you own the website. We never read a website, or let a chatbot run on it, until ownership is confirmed.

Once confirmed, we automatically read the publicly accessible pages of that website (up to 40 pages by default, only on your confirmed domain, respecting robots.txt) and use them to answer visitors. We read it again every 30 days to keep answers up to date. You can add, refresh or remove sources, and change how often they are refreshed, in your dashboard.

You decide when the chatbot starts answering visitors: it only goes live when you press Go live.

Where your plan includes a monthly message allowance, we email you when you reach about 80% and 100% of it. The chatbot keeps answering up to 120% of the allowance so your visitors are not cut off suddenly, then pauses until the allowance resets at the start of the next month or you move to a larger plan.

To protect you from unexpected costs and abuse, each chatbot also has a daily safety limit. If it is reached, visitors see a polite "temporarily unavailable" message with no mention of billing until the next day, and we email you. Visitors who send messages very quickly are asked to wait a moment.

Plans without a fixed allowance are subject to fair use. If usage is far above what is normal for your plan, we will contact you before taking any action.

By default the assistant will not state prices, availability, fees or other material property information unless it comes from data you have supplied and keep up to date.

You remain responsible for the accuracy of property and pricing information under consumer protection law, and you should review what the assistant says regularly.

Allowed by default: answering questions about vacancies and the application process, collecting contact details, and passing candidates to a person.

Not allowed by default: scoring, ranking, filtering or rejecting candidates; asking about protected characteristics; or any decision about a candidate without human review.

Collecting CVs or special category data needs a signed Data Processing Agreement, a data protection impact assessment and your own lawful basis confirmed first.

Uses that could significantly affect people, including recruitment screening, employment decisions, healthcare, credit, housing, insurance and legal advice, are switched off. They can only be enabled with our separate written approval after a review.

The assistant does not give legal, financial, medical, mortgage or other regulated advice.

Actions with significant or irreversible effects, such as payments, deletions, contracts, account changes or messages that bind you, need human approval before they happen. Agents run with the least access needed and only the permissions you grant.

Cailub supports business customers in the markets listed in the Terms and Conditions. We do not claim that the services comply with every AI law in every country. The Market Supplement for your market describes any local AI rules we account for.

The Acceptable Use Policy applies. In particular, AI services must not be used for deception or impersonation, unlawful profiling, manipulative techniques, or solely automated decisions with legal or similarly significant effects on people.

When your assistant handles personal information about your visitors, you are the controller and Cailub is your processor under the Data Processing Agreement. You must tell your visitors they are using an AI assistant and explain how their information is used in your own privacy notice.

We do not use identifiable conversations to train general-purpose AI models without a separate lawful basis and your authorisation.

You remain responsible for how you deploy and use AI services, including the decisions you make based on their outputs. Our liability is set out in the Terms and Conditions.