Procurement brief · v1 (2026-06)
Print this page →

AI Governance & Human Oversight

A one-page brief for procurement, risk, and compliance reviewers.

What Oracron does

Oracron audits freight invoices: it reads each invoice, classifies every charge, compares it to your contracted rates, and surfaces overcharges with the evidence behind them — across road, air, and sea.

How AI is used — and where the human stays in control

The AI performs detection and explanation. It does not take money-affecting action autonomously. A member of your team reviews and approves before any dispute is raised with a carrier. The AI's role is to make that decision fast and well-evidenced, not to replace it.

Why this matters for your AI Act readiness

The EU AI Act (Article 14) requires effective human oversight of high-risk AI systems, applicable from 2 August 2026. Oracron's architecture is designed around that principle:

Reviewer asks Oracron's answer
Can a human intervene before action? Yes — sign-off is required before any dispute is sent.
Are AI decisions logged? Yes — every AI call is recorded in a tamper-resistant audit log.
Can a decision be traced and explained? Yes — each finding carries its reasoning; pipeline runs are traced end-to-end.
Can a decision be reversed? Yes — nothing money-affecting executes without human approval, so it can be declined or undone before it happens.
Where is data processed? EU or US data region, chosen at onboarding.
Is customer data used to train AI? No model training on your invoice data. Anthropic and OpenAI are used under their no-training enterprise terms; prompt/response logs auto-delete within their retention windows.
Important scope note. Oracron is designed for human oversight and Article 14-ready by architecture. This brief is not a claim of formal EU AI Act certification, which is a separate regulatory process. It describes how the product is built to support your own compliance assessment.