Artificial Intelligence Transparency Notice

Last updated: 29 July 2026. Published under art. 50 of Regulation (EU) 2024/1689 (AI Act) and art. 13-14 and 22 GDPR.

In one line. AWMZA agents are software, not people. They are built on large language models, they can make mistakes, they never take a final decision about a person on their own, and your business data is not used to train anyone else's model.

1. You are talking to a machine

Every AWMZA agent that talks to a human, on WhatsApp, on a website chat, by email or by message, is an automated assistant. Where an agent is deployed on your own channels, it is set up to identify itself as an automated assistant at the start of the conversation, and a human handover is always available. As the business using the agent, you must not remove or hide that disclosure.

2. Which models we use

Our agents run on a mix of models, chosen for each task on the basis of quality, speed and cost. As of the date of this notice these are:

ProviderWhere the processing happensTypical use
Self hosted open weight models (Ollama)Our own server in Germany. Data does not leave our infrastructure.Routine classification, drafting and internal tasks
GroqUnited States, under standard contractual clausesFast conversational replies
Anthropic (Claude)United States, under standard contractual clausesLonger reasoning, writing, analysis
OpenAIUnited States, under standard contractual clausesOccasional specialised tasks

We may change providers to improve quality, reliability or privacy. Where a change means a new transfer of personal data outside the EEA, we update this page and, where required, inform customers in advance.

3. Your data and model training

We do not sell your data and we do not use the content you or your customers send through an agent to train foundation models. Our providers are engaged on business terms that exclude training on submitted content. Prompts and outputs may be retained for a short period by a provider for abuse monitoring, in line with its own policy. Self hosted models never send anything outside our server.

4. What the AI can and cannot do here

We deliberately do not offer agents for recruitment, CV screening, interview scheduling, worker evaluation, credit scoring, biometric identification or emotion recognition. These sit in the high risk or prohibited categories of the AI Act and we do not operate in them.

5. Mistakes happen, keep a human in the loop

Language models can produce content that sounds confident and is wrong, out of date, or invented. Anything that affects money, contracts, health, safety or a person's rights must be checked by a human before it is acted on or sent. You remain responsible for what you publish or send to your own customers using our tools.

6. Generated content

Text, images, posts and reports produced by an agent are AI generated. Where a platform requires machine readable marking or an explicit label for synthetic content, you are responsible for applying it when you publish. We do not create deepfakes and we do not produce content that imitates a real identifiable person.

7. Roles and responsibilities

Under the AI Act, AWMZA acts as a deployer of general purpose AI models supplied by third parties, and provides those capabilities to you inside its agents. When you switch an agent on for your own customers, you become a deployer too, with your own duties: telling people they are dealing with an AI, keeping human oversight, and having a lawful basis for the personal data you feed in. Under the GDPR you are the controller of your customers' data and we are your processor, as set out in the Terms.

8. Your rights

Anyone interacting with one of our agents can ask to speak to a human, ask what data is held about them, and ask for it to be corrected or deleted. If the agent belongs to one of our business customers, that request goes to that business. For anything about AWMZA itself, write to info@awmza.it. Further detail is in the Privacy Policy.

9. Reporting a problem

If an agent produces harmful, discriminatory, misleading or otherwise inappropriate output, report it to info@awmza.it with the conversation or a screenshot. We investigate, and where the risk is serious we suspend the agent while we do.

10. Contact

HUSSAIN MUHAMMAD MAJID — Via Abate Pelizzoni 5, 20033 Solaro (MI), Italy — VAT IT14191810960 — info@awmza.it