How we use AI — and how we keep it in check
DITAP runs a fleet of AI agents on day-to-day tasks, with human direction and accountability. This page documents the principles and safeguards behind that model for procurement and compliance teams.
See DITAP's AI-native modelPrinciples
Human supervision
Every AI agent operating client-facing or on a DITAP system has a person accountable for its performance. Supervision is not a symbolic role: it includes case review, process adjustment, and the ability to step in at any moment.
Human-in-the-loop on significant decisions
No decision with real, direct impact on a person — hiring, contract terms, billing — is made fully automatically. An agent may prepare the information and a first draft; the final decision is always made and authorized by someone on the team.
Interaction transparency
Where applicable, we make it known that an AI agent is responding. At any point in a conversation, a person can be requested, and that request is honored.
Client data doesn't train third-party models
Data a client entrusts to us is not used to train third-party AI models without a specific agreement authorizing it. The AI vendors we use are also evaluated as sub-processors (see /trust/sub-processors).
Logging and traceability
The work AI agents perform is logged: what action was taken, on what case, and what human review was involved. This allows a specific case to be audited if a client requests it or an incident requires it.
Regulatory alignment
Posture and process — we don't claim AI certifications or audits DITAP hasn't obtained.
EU AI Act — Reglamento (UE) 2024/1689
The EU's AI Act introduces transparency obligations for AI systems that interact with people. We shape our practice around those obligations — we do not claim a formal conformity certification, which the regulation does not require from a provider our size.
- We disclose when a person is interacting with an AI system rather than a human, when that interaction isn't otherwise obvious.
- We maintain human oversight over the AI systems we use, with real intervention capability — not only on paper.
- We do not use AI for practices the regulation prohibits (for example, social scoring or subliminal manipulation).
IA en reclutamiento
Applicant screening is one of the higher-risk areas for AI applied to people, and we treat it that way.
- AI-assisted screening ranks and prioritizes applications against the published role requirements — it never rejects or hires anyone on its own.
- The decision to advance, reject, or hire a candidate is always made by someone on the DITAP team.
- Any applicant can request that their case be reviewed by a person, and that review is guaranteed.
Does your procurement or legal team have questions about our AI use?
Write to us and we'll coordinate whatever detail your procurement process needs.
Email [email protected]