AI Transparency Notice

Last updated: February 24, 2026

In accordance with Regulation (EU) 2024/1689 (EU AI Act) and Article 13 GDPR (transparent information obligation).

1. Our Commitment to Transparent AI

Veluvanto uses artificial intelligence to help you manage and understand your documents. We believe you have the right to know when AI is involved, how it works, and what its limitations are. This notice explains our AI systems in plain language, as required by the EU AI Act and GDPR.

2. AI Systems We Use

Veluvanto uses the following AI-powered features:

📄 Document Analysis & Categorization

What it does: Reads your uploaded documents and automatically extracts key information — document type (invoice, contract, receipt), dates, amounts, parties involved, and other metadata. Assigns tags to organize your documents.

How it works: Uses large language model (LLM) AI to understand document content.

Limitations: AI may misclassify documents, miss or incorrectly extract data (especially from poor-quality scans or unusual document formats), or fail to process certain languages accurately. Always verify critical metadata.

🔍 Semantic Search

What it does: Allows you to search your documents using natural language (e.g., "unpaid invoices from last year").

How it works: Converts your query and document content into vector embeddings and finds the most relevant results.

Limitations: May return results that seem relevant but are not. May miss relevant documents. Not a substitute for legal document review.

💬 AI Chat Assistant

What it does: Answers your questions about your documents in natural language, with citations to source documents.

How it works: Retrieves relevant document passages and uses an LLM to generate a contextual answer (Retrieval-Augmented Generation / RAG).

Limitations: AI-generated answers may contain errors ("hallucinations"). Always verify important information against the original source documents. Do not rely solely on AI Chat for legal, financial, or medical decisions.

🌐 Document Translation

What it does: Translates your documents into other languages.

How it works: Uses LLM-based machine translation.

Limitations: Machine translation may be inaccurate, especially for legal or technical terminology. Translations are provided for informational purposes only and should not be used as certified or official translations.

📋 Document Summarization & Explanation

What it does: Creates concise summaries of your documents and explains complex or legal language in plain terms.

How it works: Uses LLM to analyze and restate document content.

Limitations: Summaries may omit important details or nuances. Do not replace legal counsel with AI-generated explanations of contracts or legal documents.

🔔 Automated Reminders

What it does: Detects important dates (invoice due dates, contract expiry) and suggests reminders.

How it works: AI extracts dates from documents and matches them to contextual patterns (e.g., "due date" field).

Limitations: AI may miss dates or create incorrect reminders. You are responsible for confirming and managing your own deadlines.

3. AI Risk Classification (EU AI Act)

Under the EU AI Act (Regulation (EU) 2024/1689), AI systems are classified by risk level.

Veluvanto AI classification: Limited risk under Article 50 EU AI Act

Our AI systems do not fall into the high-risk categories listed in Annex III of the EU AI Act (such as employment screening, creditworthiness assessment, biometric identification, or safety-critical systems). Our AI is used for document organization and information retrieval to assist users — not to make consequential decisions about people.

We comply with the transparency obligations applicable to limited-risk AI systems under Article 50 of the EU AI Act, including the obligation to inform users when they are interacting with an AI system where this is not otherwise obvious.

4. No AI Training on Your Data

Your documents are never used to train AI models — neither our own models nor those of our AI providers. Your data is processed only to provide the service to you. Our AI provider agreements include explicit data processing restrictions and prohibitions on using your data for model training.

5. Human Oversight and Your Control

We maintain meaningful human oversight of our AI systems:

  • All AI-generated metadata (tags, entities, dates) can be reviewed and corrected by you at any time
  • AI-suggested reminders require your explicit confirmation before activation
  • AI Chat answers include citations so you can verify information in source documents
  • No AI-generated output has binding legal or financial effects without your review
  • You can delete your documents and data at any time (see Privacy Policy)

6. AI Providers and Third Parties

We use the following third-party AI infrastructure (acting as our data processors):

  • Google LLC — Vertex AI (Gemini models) — Large language model inference for document analysis, semantic search, AI chat, translation, and summarization. Processing location: EU Google Cloud regions (europe-west — document content does not leave the EEA). Subject to Google Cloud Data Processing Addendum. API data is not used for model training by default under Vertex AI terms. See Google Cloud data governance for details.

All AI providers are bound by data processing agreements ensuring compliance with GDPR and EU AI Act requirements.

7. Accuracy, Errors, and Limitations

AI systems are not perfect. You should be aware of the following:

  • Hallucinations: AI may generate plausible-sounding but incorrect information. Always verify AI-generated answers against original documents.
  • OCR limitations: Poor-quality scans or handwritten documents may be extracted inaccurately.
  • Language limitations: AI performs best in commonly used languages; less common languages may have lower accuracy.
  • Not legal/financial advice: Nothing generated by Veluvanto AI constitutes legal, financial, tax, or medical advice.

8. Automated Decision-Making (GDPR Art. 22)

Veluvanto does not make solely automated decisions that produce legal effects or similarly significantly affect you as required to be declared under Article 22 GDPR.

All AI outputs (metadata, tags, summaries, reminders) are informational and subject to your review and override. You remain in control of all decisions based on AI-generated information.

9. Your Rights Regarding AI Processing

In relation to AI processing of your data, you have the right to:

  • Correct any AI-generated metadata directly in the application
  • Request human review of any AI-generated output by contacting us
  • Request deletion of your documents and all derived AI-generated data
  • Object to specific types of AI processing (contact us at privacy@veluvanto.com)
  • Receive an explanation of how our AI systems work

10. Feedback and Complaints

If you have concerns about our AI systems, believe an AI output has harmed you, or wish to report a potential AI system issue:

Contact us at ai@veluvanto.com or legal@veluvanto.com. We take AI safety feedback seriously and will respond without undue delay.

You may also contact the relevant national AI supervisory authority. In the Czech Republic, oversight of EU AI Act compliance is coordinated by:

  • Czech Telecommunication Office (ČTÚ) — national market surveillance authority and main point of contact for AI Act; website: www.ctu.cz
  • ÚOOÚ — Úřad pro ochranu osobních údajů — supervisory authority for data protection aspects of AI processing; website: www.uoou.cz

Note: The Czech national AI Act implementation law (amending Act) is expected to enter into force in August 2026. Authority designations may be updated as the framework is finalised.

11. Updates to This Notice

We will update this notice as our AI systems evolve or as regulatory requirements change. Material changes will be communicated via email. We aim to maintain full transparency about our use of AI at all times.

12. Contact

For questions about our AI systems or this notice:

Veluvanto s.r.o.
AI inquiries: ai@veluvanto.com
Legal: legal@veluvanto.com