1. Who and what for
We use AI in our own work and in our work for clients. This policy is maintained by Helobrand sp. z o.o., Kraków, which operates skrits. It supplements the privacy policy and the terms; it does not replace them. It describes how we work. It is not a certificate or legal advice.
2. How we use AI
AI helps us with research, with drafts of texts and analyses, with summaries, with sorting mail and tasks, and with building tools and automations. We show examples on the Proof page. AI produces the first version. It does not replace our knowledge or our responsibility.
3. A human decides
Everything that leaves the company is prepared by AI and approved and sent by a human: offers, prices, contracts, messages to clients and publications. The person reviewing opens the sources and verifies numbers, names and dates. We check claims from books against the original. Important decisions, strategies and plans go through our models and guidelines before we approve them.
4. No automated decisions about people
AI does not make decisions about people here. It does not assess, classify or filter candidates or employees, does not monitor teams, and does not decide whether to accept an enquiry, how to price it, whether to start working together, or who gets access to services. A human makes or approves such decisions.
5. Data and confidentiality
We work in tools on company accounts, never on private free accounts. We do not paste into AI: passwords and login credentials, full card numbers, account numbers and IBANs, people’s tax and identification numbers, health data, or personal data of clients, candidates and employees. For each task we provide only as much data as is needed. Client materials stay on the client’s accounts; we work on them under a data processing agreement and a confidentiality agreement. Details on personal data are in the privacy policy.
6. Tools
We use tools that have been checked for their data processing terms, including Claude (Anthropic), the OpenAI API, Google Workspace with its Gemini features, and Cloudflare. According to Google’s declaration, Gemini features in Workspace do not use the content of our accounts to train models. A new tool goes through our review before it touches company or client data. We record and transcribe conversations only with the consent of all participants.
7. Agents and permissions
No AI gets access to mail, drive, a client’s system, a bank or social media accounts without written consent, a defined purpose and the least access needed. By default, access is read-only. AI does not send, pay, publish or delete on its own. Every change made by AI is labelled and logged. We do not combine private data, outside content and the ability to send data outside in a single agent.
8. We tell you about AI
In the materials we deliver, we say when AI helped with them and that a human reviewed them. If we launch a conversation with an AI assistant, we tell you at the very start. We label realistic images, recordings or audio generated by AI. In graphic design projects, we say which elements are generated.
9. Others’ rights and integrity
AI-generated content is not automatically free of third-party rights. Before use, we check copyright, image rights, trademarks and secrets. We do not use AI to mislead, to impersonate anyone, to discriminate, or to create fake reviews, recommendations or references.
10. Errors and incidents
An AI error that left the company, or nearly did, is reported internally within 1 working day, without hunting for someone to blame, and entered in a register. If an error or incident affects your data or work, we tell you without undue delay. We assess and report personal data breaches in line with the GDPR.
11. Training and review
People who use AI in our work are trained and entered in a training register. The Regulation on artificial intelligence (the AI Act) requires companies to take measures supporting the AI competence of the people who use it, and that is what we document. We test repeatable AI processes on sets of cases before they touch a client’s work. We review this policy every quarter and whenever tools or regulations change.
12. Legal compliance
We apply the rules on data protection (the GDPR) and on artificial intelligence. We do not claim that our work is “AI Act compliant”; we check the binding wording of the law with a lawyer before we use it in an offer to a client.
13. Questions and changes
Questions, comments and reports: hi@skrits.com. The current version is always on this page, with its date. The English version of this policy is a courtesy translation; in case of any discrepancy the Polish version prevails.