An AI compliance agent for an online store checks every day the things that could bring you before ANPC or an unhappy customer: withdrawal and refund deadlines, the mandatory information on product pages, the reference price for discounts, personal data requests and the status of invoices in e-Factura. It flags and prepares, and legal interpretation stays with a person.


An online store is subject to more rules than it seems. Customers have the right to withdraw from the contract within 14 days, under OUG 34/2014, and the money has to be returned within a fixed deadline. Discounts have to be shown against the lowest price charged in the last 30 days, under the rule introduced by the Omnibus directive. Product pages must contain the manufacturer information required by the EU General Product Safety Regulation. Customer data falls under GDPR. None of this is hard on its own. What’s hard is tracking all of it, every day, across hundreds of products and orders.
Counts the legal deadlines on every return and refund request and alerts you before they run out
Checks product pages for missing information: manufacturer, warnings, instructions in Romanian, warranty
Compares the crossed out price in each promotion with the product’s price history and flags discounts that are hard to justify
Keeps the log of requests to access, correct or delete personal data and the response deadline for each one
Watches SPV for invoices rejected or not transmitted in e-Factura and puts them on the list to fix
The AI agent doesn’t tell you what the law says in your case. It shows you where the store’s day to day practice drifts from the rules you set together with your lawyer or consultant.
Tell us what slows you down. We talk, then decide together whether and how we work.

Every morning you get a short list of deadlines. Which returns must be refunded by the end of the week, which GDPR request has a deadline coming up, which invoice was rejected in e-Factura and needs to be resent. The list is sorted by due date, not by the order in which requests came in.
Before every campaign, the AI agent goes through the products included. For each discount it checks the reference price in the platform’s history and flags the products where the crossed out price doesn’t appear in the last 30 days. You decide whether to adjust the displayed price or pull the product from the campaign. For eMAG Marketplace it also checks the platform’s listing rules.
Once a month, it reviews the new products added to the catalog and the terms and conditions page, the return policy and the privacy policy. If you changed courier, delivery time or payment processor and the pages still describe the old setup, it shows you the difference. A person approves the new text, ideally after the lawyer has seen it.
Also monthly, the AI agent compares the subscriber list in the email tool with the recorded consents. If someone asked to unsubscribe by replying to an email or by messaging support, and the address is still active on the list, you see it in the report. Contacts imported without a clear source of consent show up too. A person decides what to do with them.

Gomag, Shopify, WooCommerce, PrestaShop or Magento, for product pages, price history and legal pages
eMAG Marketplace, for the listings there and the platform’s rules
The returns flow in your platform or helpdesk, Freshdesk, Zendesk or Gorgias, for customer requests
SPV ANAF and your invoicing software, SmartBill, Oblio or FGO, for invoice status in e-Factura
Klaviyo, Mailchimp or Brevo, for marketing consent and unsubscribes
Google Drive, where you keep the approved policies and the GDPR request log
The AI agent starts from a set of rules written for your store: the deadlines you apply, the mandatory information by product category, who answers for what. A person with legal training validates the rules. The AI agent applies them consistently, it doesn’t invent them.

It doesn’t give legal advice and it doesn’t replace your lawyer or GDPR consultant. It doesn’t guarantee that the store is compliant. It shows you deviations from the rules you set, but the rules have to be right from the start.
It doesn’t respond to authorities and it doesn’t sign anything on behalf of the business. It doesn’t refuse returns on its own and it doesn’t decide that a customer’s request is abusive. And it doesn’t delete personal data without approval, however clear the request may seem.

We start from what already exists: the return policy, the terms and conditions, the privacy policy and the way you handle returns and data requests today. If you work with a lawyer, it helps to have them involved in setting the rules. The AI agent can’t be better than the rules it receives.
The first step is usually tracking return and refund deadlines, because a missed deadline shows up right away there. Then we add product page checks, pre-campaign discount checks and the GDPR log. You learn the cost of each tool upfront.
Anything that means legal interpretation stays with people. If a customer disputes a refused return, if you get a complaint from ANPC or a letter from a lawyer, the AI agent pulls together the whole file: the order, the correspondence, the photos, the deadlines, the policy in force on the date of sale. A person writes or approves the reply, and in serious cases a lawyer reviews it.
For GDPR requests, the AI agent finds the customer’s data in the platform, the helpdesk and the email tool and prepares the response. The actual deletion and the reply sent to the customer are approved by the person responsible for data protection in the business. A wrong deletion can’t be undone.
Changes to legal pages and to product pages flagged as incomplete also go through a person. The AI agent suggests the missing text from the supplier’s data, but it doesn’t publish warnings or safety information on its own.
The most important link on the page. From here you go up to the agent we sell.
Sideways, on the same level, so you can move easily between pages.
Not in a legal sense. It shows you where daily practice drifts from the rules set with your lawyer or consultant: a refund deadline about to be missed, a product page without manufacturer details, a discount calculated against a price that doesn’t appear in the history. The legal conclusion is drawn by a person qualified to do so.
For every withdrawal request, the AI agent records the date, calculates the deadline by which the money must be returned and alerts you in good time. If the product hasn’t come back yet or the customer hasn’t sent proof of shipment, it notes that too. A person approves and makes the actual refund.
Yes, if the platform keeps a price history. For each product in the campaign, it compares the crossed out price with the lowest price charged in the last 30 days and flags the differences. It doesn’t change prices on the site. It gives you the list, and you decide what to fix before the campaign starts.
It logs the request and the response deadline, looks for the customer’s data in the platform, the helpdesk and the email tool and prepares a list of what can be deleted and what must be kept, for example the invoices that accounting law requires you to retain. The deletion and the reply to the customer are approved by the responsible person in the business.
It quickly gathers everything related to the case: the order, the correspondence with the customer, the photos, the deadlines, the version of the return policy valid on the date of sale. It puts the file in chronological order. The reply to ANPC is written and signed by a person in the business, ideally after a lawyer has read it.
Tell us what slows you down. We talk, then decide together whether and how we work.