The EU Artificial Intelligence Act (AI Act) entered into application in August 2026. This page describes how Nepteo complies with its obligations.
1. Our approach
Nepteo is an AI-powered workspace for independent professionals. It uses third-party large language models (Anthropic Claude, OpenAI GPT, Google Gemini) through their commercial APIs to process user requests, manage business data, and automate administrative tasks.
We build on top of general-purpose AI models. We do not develop or train foundation models ourselves.
2. AI Act classification
Under the AI Act risk-based classification framework, Nepteo operates as a deployer of general-purpose AI systems, not as a provider of foundation models. Our assessment:
- Not high-risk: Nepteo does not make decisions in areas classified as high-risk by the AI Act (employment, credit scoring, law enforcement, critical infrastructure, education, access to essential services). It is a productivity tool for business administration.
- Transparency obligations apply: as a system that interacts with users through natural language, Nepteo must clearly inform users that they are interacting with an AI system.
- General-purpose AI obligations: are the responsibility of our upstream model providers (Anthropic, OpenAI, Google). We monitor their compliance disclosures.
Nepteo uses third-party AI models (Anthropic, OpenAI, Google) to generate responses. Nepteo does not use your data to train AI models. What each provider's own terms say is on the sub-processors page.
3. Transparency
Nepteo is transparent about AI involvement:
- Every response from the AI is clearly identified with the Nepty label and icon.
- Actions taken autonomously by the AI (scheduling, sending, updating) are logged in an audit trail visible to the user.
- The AI discloses when it cannot complete a request or when a feature is not available.
- Users can review, approve, or reject AI-suggested actions before they take effect for sensitive operations.
4. Human oversight
Nepteo provides meaningful human oversight over AI decisions:
- Approval gates: sensitive actions (sending invoices, emails on the user's behalf) require explicit user confirmation before execution.
- User-defined rules: the user defines rules and boundaries that the AI must respect. The AI cannot override these rules.
- Audit log: every action performed by the AI is logged and reviewable.
- Override: you can correct or undo the actions Nepty proposes, at any time.
5. Data protection
AI-specific data protection measures:
- Nepteo never uses your data to train AI models (see what each provider's terms say).
- Persistent memory (PSM) is scoped per user and never shared across accounts.
- Users can view, edit, and delete any fact stored in their AI memory.
- Memory data is included in data export (portability) and deletion requests.
6. Contact
For questions about our AI compliance practices: privacy@nepteo.ai.