Responsible AI Statement
Last updated: 15 June 2026
Corpus Logic is built around an AI worker, NARUS, that reads operational data, reasons over your business, and prepares actions for your team. This statement explains how we design and govern that capability so it stays useful, accountable, and safe. It complements our Privacy Policy and Acceptable Use Policy.
1. Our core principle: humans at the commit
NARUS is designed to be maximally helpful up to the decision point and deliberately conservative at it. It perceives, remembers, reasons, and drafts — but a person approves every outward or irreversible action. Intelligence assists the decision; it does not replace the decision-maker.
2. What the AI does
- Reads incoming operational signals such as emails, documents, and live position data.
- Extracts structured information and locates decision-bearing figures from source.
- Reasons over the wider context — jobs, lanes, fleet, and history — to surface what matters.
- Prepares drafts: quotes, replies, route plans, and recovery actions, each with rationale.
3. What the AI does not do
- It does not book carriers, send customer commitments, or file customs declarations automatically.
- It does not take irreversible commercial actions without explicit human approval.
- It does not provide legal, regulatory, tax, or financial advice — compliance content is information only, with links to official sources.
- It does not use one customer’s data to train models for another customer, and it never crosses tenant boundaries.
4. Human oversight and control
Material actions arrive as drafts for review. Your team can approve, reject, or override each one — typically in a single click — with the reasoning and source figures attached. Where we offer configurable autonomy for low-risk, reversible steps, it defaults to the most conservative setting and can only be raised deliberately by an authorised user.
5. Accuracy and its limits
Decision-bearing values such as rates, weights, and dates are located verbatim from source rather than paraphrased, and material actions hold for human review if a required figure cannot be confidently resolved. Even so, AI outputs can be incomplete or wrong. You remain responsible for decisions made using the platform, and you should verify outputs before relying on them for production or regulated activities.
6. Data use and model providers
Content sent to AI models is limited to what the task requires and is processed by the providers listed on our Sub-processors page. We do not sell your data, and we do not use one customer's operational data to improve another customer's outcomes. Personal data handling is governed by our Privacy Policy and, for customers, Data Processing Agreement.
7. Transparency and audit
Material AI-worker actions are recorded in an append-only audit trail — what was received, how it was reasoned, what was drafted, and what a person approved. This supports accountability, dispute resolution, and your own governance obligations.
8. Fairness and appropriate use
The platform must not be used to make decisions that unlawfully discriminate, or to generate deceptive or harmful content, as set out in our Acceptable Use Policy. We design features to keep a human accountable for consequential outcomes.
9. Raising a concern
If you have a concern about an AI output relating to you, or about how the AI features behave, contact support@corpuslogic.com (data matters) or support@corpuslogic.com (product matters). We take such reports seriously and will respond promptly.