Use of Artificial Intelligence

1. What the AI does#

• Replies to your customers 24 hours a day, in your company's tone, based on the instructions and information you entered in the panel. Available from the Essential plan.
• Understands the audio messages customers send, transcribing them before replying.
• On the Professional plan, it checks your prices and your schedule to give quotes and book appointments.
• Hands off to the team when the rule you defined says to call a human.

The cost of the AI on the plans that include it is borne by atendIA: you do not create an account with an AI provider and you do not pay per message answered.

2. What is sent, and to whom#

To generate a reply, the content of the conversation is sent to the model provider chosen in your company's configuration (OpenAI, Anthropic, or Google). Received audio goes to transcription. Details on each provider are in Sub-processors.

• When the AI checks the schedule or services, what goes back to the model is time, service, and duration. The contact's name and phone number are left out, by design, verified by an automated test.
• With the AI turned off, no conversation content is sent to a model provider.

3. What we do not do#

• We do not train models on your conversations, and we contract providers under terms that do not use content sent through the API to train their models.
• We do not sell or share conversation content for advertising.
• We do not use one customer company to teach another's AI: each configuration belongs only to your company.

4. The limits of AI#

Every AI system makes mistakes. It can misunderstand a question, confidently answer something that is not in your information, or be fooled by a message designed for that. atendIA reduces this risk with the instructions you define and the hand-off rule to the team, but does not eliminate it.

• The AI does not replace a licensed professional: it gives no diagnosis, legal opinion, or investment recommendation.
• atendIA does not guarantee any commercial result from the AI.
• Responsibility for what is promised to your customer lies with your company, as stated in the Terms of Use, section 9.

5. What falls to whoever configures it#

• Correct information. The prices, hours, address, and rules the AI uses are the ones you entered. Keep them up to date.
• Review. Monitor what the AI is replying, especially in the first weeks and after changing the instructions.
• Transparency with your customer. We recommend that the service make clear that a virtual assistant exists, and the AI cannot deny being an AI when the customer asks directly.
• Human review of decisions. If your company uses the AI for decisions that affect the customer (approving, refusing, classifying), ensure review by a person when the customer requests it, as provided in art. 20 of the LGPD (Lei nº 13.709/2018, Brazil's data protection law).

6. Prohibited uses of the AI#

• Pretending to be a specific real person, or misleading the customer about who is replying.
• Pressuring, manipulating, or exploiting a customer's vulnerability to close a sale.
• Asking for sensitive data that is not necessary for the service.
• Any use prohibited by the Acceptable Use Policy.

7. How to turn it off or adjust it#

The AI is turned on and off in the panel, for the whole company. Your team can take over any conversation with one click, at any time. Questions about how the AI handles data: [email protected].