Trust & Policies · AI Use Policy

How Plutonia Uses AI — and How It Is Constrained

Plutonia uses AI-assisted workflows to accelerate procurement work, with people reviewing findings and making the final buyer-facing recommendation. This policy explains what AI does, what it does not do, the governance model, and the controls that keep outputs honest and confidential. Last updated June 2026.

Scope

What This Policy Covers

This policy applies to AI-assisted workflows Plutonia uses in supplier discovery, quotation analysis, document organization, risk flagging, quality-control coordination, and buyer reporting. It does not make Plutonia an autonomous platform: people remain responsible for buyer-facing recommendations and decisions.

What AI does: accelerate research, extract and normalize information, organize and tag documents, surface possible risk flags, and prepare first drafts of reports.

What AI does not do: approve suppliers, release payments, sign off on compliance, certify products, or make final procurement decisions.

Governance

A Govern, Map, Measure, Manage Approach

Plutonia's approach to AI governance is informed by recognized AI risk-management principles, applied across the workflow lifecycle rather than as a one-time setup.

Controls

Minimum Governance Controls

🔒

Human approval on shortlists, risk flags, and final reports

Prevents over-reliance and false certainty.

“Final buyer-facing recommendations are human-reviewed.”

🔒

Source logging on every report section

Enables provenance and client challenge or review.

“Each report section can link to its underlying source material or operational note.”

🔒

Prompt and data segregation by client

Prevents leakage across projects.

“Confidential buyer and supplier information is handled within controlled workflows.”

🔒

Redaction and access control

Reduces privacy and commercial-confidentiality risk.

“Sensitive files are shared only on a need-to-know basis.”

🔒

Defined meaning of risk scores

Prevents legal or compliance overclaiming.

“Risk scores are internal assessments, not legal determinations.”

🔒

Content and claim review workflow

Keeps the website aligned with evidence.

“Pages are reviewed and updated on a defined cadence.”

Known risks we manage

AI Risks We Actively Manage

Any AI workflow touching supplier documents, contract terms, or buyer data carries risk. We manage these with retrieval guardrails, output filtering, access control, audit logs, version traceability, and explicit human approval gates.

Related

Related Trust Pages

FAQ

Frequently Asked Questions

How does Plutonia use AI?
Plutonia uses AI-assisted workflows to accelerate supplier discovery, RFQ extraction, quote normalization, document organization, risk flagging, QC coordination, and first-draft reporting. People review findings and make the final buyer-facing recommendation.
Does AI approve suppliers or release payments at Plutonia?
No. AI does not approve suppliers, release payments, sign off on compliance, or make final procurement decisions. Those are human decisions.
What does a risk score mean?
A risk score is an internal, evidence-based assessment used to prioritize attention. It is not a legal determination or a guarantee.
How is confidential information protected?
Through prompt and data segregation by client, redaction, access control, and human review, so confidential buyer and supplier information stays within controlled workflows.
How does Plutonia keep AI outputs accurate?
Through human review, source logging on report sections, defined meanings for scores, and a content and claim review workflow on a defined cadence.
Does Plutonia use AI-generated media on trust-critical pages?
Where AI-generated or heavily edited media is used, Plutonia avoids it on trust-critical pages or labels it clearly.

Questions About How We Use AI?

We are happy to explain our methods, limits, and controls before you engage.