VeillantVeillantBack to home

AI Policy

Effective as of: August 21, 2026

This AI Policy applies to your access and use of Veillant's AI features and forms part of our Terms of Service. Capitalized terms not defined here have the meaning given in the Terms.

1. What "Veillant AI" means

"Veillant AI" means the tools and functionality that use machine-learning and artificial-intelligence models to power your voice agents, including understanding callers, generating natural real-time speech and text responses, and extracting information from the documents and configuration you provide.

2. Input and output

You may provide input (the agent's configuration, knowledge, and the caller's speech during a call, "Input") and receive output generated by the AI ("Output"). Input and Output are your Customer Content. You are solely responsible for your Input and Output and for ensuring they do not violate any law or third-party right. Because of the nature of machine learning, Output may not be unique and similar Input may produce similar Output for different users.

3. Real-time processing & end-of-call analysis

Call audio is processed in real time to enable the conversation and is not recorded or stored by Veillant. The transcript exists only transiently, in memory, so the AI can respond, and is not persisted by us. For every account, an AI step reads that in-memory transcript once at the end of the call and derives only what fits a closed menu of categories (the reason for the call, its outcome), then discards the transcript. The result is stored as decoupled, non-identifying counters. See our Privacy Policy for details.

4. Training & system improvement

We do not use your Customer Content to train our own models, and we select AI providers that, by default, do not use data sent through their API to train their models. Your use of Veillant AI does not grant us any right to use or share your Customer Content beyond what is needed to provide the Service. We do use anonymous, aggregated signals, with no link to any account or caller, to improve the quality and reliability of the Service and to understand demand at a coarse level (for example, which regions and hours generate calls). Because these signals are not personal data, they do not identify you or your callers.

5. Sensitive & regulated data

Do not configure your agent to ask for, or to collect, sensitive or regulated information (protected health information, government-issued identification numbers, financial-account credentials) unless you have a separate written agreement with us permitting it. This is about what you design your agent to collect: callers can always volunteer whatever they choose, and Section 3 of our Data Processing Agreement describes what happens then and what you, as controller, need to have in place. Veillant AI is not intended for use in medical diagnosis, treatment, or other regulated professional activities, and Output is not a substitute for a licensed professional.

6. Acceptable use

You may not use Veillant AI or any Output to:

  • Develop competing AI models, or reverse engineer or extract our models or systems.
  • Deceive people about the AI nature of the agent where disclosure is required, or impersonate others.
  • Generate spam, fraudulent, harassing, or unlawful content, or place unsolicited or abusive calls.
  • Violate any applicable law, regulation, or third-party provider policy.

You are responsible for disclosing the AI nature of your agent and obtaining any consents required in your jurisdiction.

7. Third-party AI providers

Veillant AI incorporates models from third-party providers (currently OpenAI). We may add, remove or change providers to improve quality and reliability. Where a new provider would process the data of the people who call your assistant, we publish it on our Subprocessors page and notify registered customers at least 30 days before it starts processing, and you may object, as set out in Section 5 of our Data Processing Agreement. Their processing is subject to their own policies.

8. Disclaimer

VEILLANT MAKES NO WARRANTY AS TO THE RESULTS OBTAINED FROM VEILLANT AI OR THE ACCURACY OF ANY OUTPUT, WHICH IS PROVIDED "AS IS." OUTPUT MAY CONTAIN INACCURACIES; YOU SHOULD INDEPENDENTLY VERIFY IT BEFORE RELYING ON IT, AND MUST NOT USE IT AS PROFESSIONAL, LEGAL, MEDICAL, OR FINANCIAL ADVICE. We may modify or discontinue AI features at any time.

9. Contact

Questions about our use of AI? Contact us at hello@veillant.eu.

Legal