Version 1.0. This policy explains where we use artificial intelligence in the Euro Intermed B2B platform and across the ecosystem sites (Angrosist, PalletClearance, SkalYou), what it does and does not do, and how you can recognise content produced with AI. It is published under the transparency obligations of Article 50 of Regulation (EU) 2024/1689 (the AI Act), applicable from 2 August 2026, and complements the Privacy Policy, which remains the reference document for personal data processing.
1. The underlying principle
We neither hide our use of AI nor overstate it. Three rules govern everything below:
- When you are talking to an automated system, you are told so in the first message. We do not leave you with the impression that you are talking to a person.
- AI does not make commercial decisions on a person’s behalf. It does not approve, reject, set prices or select partners.
- You always have access to a person. WhatsApp, email or phone, without going through the assistant.
2. The conversational assistant (“Bogdan”)
The chat widget on our sites and our WhatsApp conversations are operated by an automated assistant powered by artificial intelligence. It identifies itself as such in the first message of every conversation, on every channel.
What it does: it runs the qualification conversation — what you are buying or selling, what quantity, where it delivers, your company details — and verifies the company registration number against the public register.
What it does not do: it does not negotiate, confirm prices, accept orders, conclude contracts, or decide whether a request is accepted. All of that stays with the human team.
How to reach a person: type “operator” (or “I want to speak to someone”) in the conversation, use the human-contact button in the chat window, or go straight to the details on the contact page. When a conversation moves to a colleague you are told explicitly; we do not take over silently.
What happens to what you write: messages are sent to a language-model provider to generate the reply. The model receives the conversation text; it does not receive direct access to our database, to files, or to credentials. When a real operation is needed — verifying a company number, saving a request, transferring to a person — the model only proposes the action, and our own code validates and executes it. Details of providers, legal bases and transfers are in the Privacy Policy, sections 6 and 8.
We do not make automated decisions producing legal effects concerning you within the meaning of Art. 22 GDPR. See section 11 of the Privacy Policy.
3. Where else we use AI
| Where | What it does | Human oversight |
|---|---|---|
| Published offer titles | Normalises a product’s commercial name into a readable, consistent title | The house style is set by us; a person publishes the offer |
| Supplier offer intake | Extracts products, quantities and prices from Excel, PDF or Word files sent by suppliers | The output is checked before publication |
| Category classification | Picks a category from a closed list defined by us | The model cannot invent categories; it picks or declines |
| Article drafting | Research and structuring | See the Editorial policy |
4. AI-generated content: what we label and what we do not
We label any blog article that was generated automatically and did not go through human review. We do not publish such articles today — every article goes through the process described in the Editorial policy, which is documented, attributed and verifiable. If that changes, the label appears on the article, visibly, not buried in terms.
We do not label commercial text — product titles, category descriptions, service pages. Article 50 covers text published to inform the public about matters of public interest (legislation, politics, economics, health, education, justice, public safety, the activity of public authorities). A wholesale product title is not in that category. The reasoning is recorded in our internal AI systems register, not only here.
Images, video and audio. We do not publish AI-generated images, video or synthetic voices that could be mistaken for real recordings. Product and category photographs are real or licensed images. We use no synthetic human models, no cloned voices and no generated lip-sync. Should we ever use such material, it will be marked as required by Art. 50(4).
5. What we do not do
- We do not use AI to assess individuals — no credit scoring, no candidate screening, no decisions about access to services based on an automatically generated profile.
- We do not publish automatically, without review, on sensitive subjects (health, financial, legal).
- We do not use customer conversations to train our own models.
- We do not present a model’s output as a commercial confirmation.
6. Our role and our responsibility
For the purposes of the AI Act, EURO INTERMED SOLUTIONS S.R.L. is a deployer of the AI systems described above: we use them in the course of our professional activity, but we neither develop them nor place them on the market under our own name. The language models belong to their providers.
We maintain an internal register of the AI systems in use, recording each one’s purpose, the data entered, whether its output is published, and the applicable obligation or the reason for exemption. A person responsible for AI compliance is designated internally and can be reached at the address below.
7. Errors and complaints
AI systems get things wrong. If the assistant told you something incorrect, if published content looks automatically generated and unlabelled, or if you want to contest a processing operation, write to contact@euro-intermed.com. We will respond and correct it.
8. Changes
This policy is updated when the systems we use or the applicable legal framework change. The date of the last update is shown at the top of the page.