Responsible AI

Responsible AI, by design

Our agents are built with GDPR and the AI Act in mind. Here is what that means for your company and what you get from us.

Where the agents run

01

Infrastructure chosen together

We pick providers and hosting locations to match your data, budget and sector requirements.

02

EU hosting on request

When your data calls for it, the agents run on infrastructure in the European Union.

03

On your infrastructure

If you already have servers or a cloud, we can deploy and configure the agents there.

GDPR: the question and our answer

Data transfers outside the EU (GDPR Chapter V).

We choose providers and hosting regions that fit your data; EU hosting on request and contractual clauses where needed.

Every AI provider used becomes a processor.

We tell you in writing which providers each agent uses and what data they receive.

Processor contract (Art. 28).

Data Processing Agreement included with every project.

Security of processing (Art. 32).

Access control, encryption and logs for every agent.

Impact assessments (Art. 35).

DPIA support included in the AI audit.

Automated decisions affecting people (Art. 22).

Human approval steps built in where decisions affect individuals.

Erasure and minimisation (Art. 17, Art. 5).

We agree up front what data the agent uses and how long it keeps it.

GDPR fines reach €20 million or 4% of global turnover. The Romanian authority is ANSPDCP.

EU AI Act: the timeline

  1. 1 Aug 2024Regulation (EU) 2024/1689 enters into force.
  2. 2 Feb 2025Banned practices; AI literacy duty (Art. 4).
  3. 2 Aug 2025Rules for general-purpose AI model providers; penalties apply.
  4. 2 Aug 2026Transparency duties (Art. 50).
  5. 2 Aug 2027High-risk rules for AI in regulated products.

Where our agents fall

Support and Sales agents are limited risk (transparency duties). Finance and Operations agents are minimal risk in normal use. Uses such as screening job candidates or credit scoring can be high-risk; we classify each use case in the audit.

What we provide

  • AI literacy training
  • Built-in AI disclosure in Romanian and English
  • Written risk classification
  • Human oversight
  • Logs for every agent
  • We tell you before changing the model an agent uses
  • A compliance pack per agent

Fines reach €35 million or 7% of turnover for banned practices and €15 million or 3% for most other breaches, with lower caps for SMEs.

Sector rules

01

NIS2

In Romania: GEO 155/2024, authority DNSC. Companies in transport, logistics, energy, health, manufacturing, food and digital services must vet supplier security. We give you the documentation you need.

02

DORA

Since 17 January 2025, financial firms must manage ICT third-party risk. We document the providers and dependencies of each agent.

03

Data Act

Since 12 September 2025, switching cloud providers is easier. We build agents so they can be moved.

Our commitments

  • A data processing agreement with every project
  • A written list of the providers each agent uses
  • You remain data controller and AI deployer
  • Designed for GDPR and the AI Act; helps you meet your obligations
  • Concrete security measures, described in writing
Let's talk

Questions about AI and data? Talk to us.