AI Transparency

Our approach

Halconia treats AI as business infrastructure. Technology is the means: first we understand the process, the data, the constraints and the level of human oversight required.

Where we use AI

• Internal agents trained on knowledge specific to the organisation.
• Processing and classification of documents, emails or operational information.
• Decision support with contextualised and traceable information.
• Automations that remove repetitive tasks under agreed rules.

Human control and data

We define with the client what is automated, what is reviewed and what always remains a human decision. We prioritise data minimisation, appropriate permissions, traceability and protection of business knowledge.

European Artificial Intelligence Regulation

The EU AI Act (Regulation (EU) 2024/1689), laying down harmonised rules on artificial intelligence, classifies AI systems according to the risk they pose to people's health, safety and fundamental rights. Halconia adopts that framework as a design criterion, regardless of the role it plays in each project as a provider or as a deployer.

Classification by risk level

Before deploying a system we assess with the client which category it falls into and what obligations arise from that:

• Unacceptable risk: we do not design or deploy systems included among the practices prohibited by article 5 of the AI Act, such as social scoring, subliminal manipulation or emotion recognition in the workplace.
• High risk: where a use case may fall within Annex III of the AI Act — for example, recruitment processes, access to essential services or management of critical infrastructure — we expressly flag this and only proceed if it is possible to meet the requirements on risk management, data quality, technical documentation, activity logging, human oversight and robustness that the rules demand.
• Limited risk: in systems that interact with people or generate content we establish transparency obligations, so that it is clear that the user is interacting with an AI system and that generated content is identifiable as such.
• Minimal risk: most of the internal automations we deploy fall into this category; even so, we apply our own traceability and review criteria to them.

Meaningful human oversight

No system we deploy replaces human judgement in significant decisions. For each project we define in writing who oversees the system, what information they have in order to understand its outputs and limitations, and how they can intervene in, correct or stop its operation. Oversight must be effective and not merely formal: it requires people with knowledge of the process, a genuine ability to review and records that make it possible to reconstruct how a result was reached.

We likewise document the known limitations of each system, including the possibility of incorrect outputs, and we train the teams that will use it, in line with the duty of AI literacy set out in article 4 of the AI Act.

On this website

This site does not currently use AI chatbots or automated support. Enquiries received through the form are reviewed by members of the team.

Through this website no decisions are taken based solely on automated processing, including profiling, which produce legal effects concerning individuals or similarly significantly affect them, within the meaning of article 22 of the GDPR. Nor is the information received used to automatically classify those who write to us. You can find details of our processing operations in our Privacy Policy.

If in future we incorporate any AI-based functionality into this site, we will state this clearly before it is used and we will update this page. For any enquiry about our responsible AI approach you can write to us at hola@halconia.es.