AI plays by the same rules as your people.
Important decisions benefit from analysis, pattern recognition, and contextual retrieval. Advisory Intelligence brings these capabilities into the governed decision process as bounded participants: attributed in the record, reviewable in place, operating within the same authority that governs human contribution.
AI contribution inside the governed decision, not alongside it
Advisory Intelligence is a set of capabilities that each participate within the governed decision process. Each has a defined scope, an attribution model, and graceful continuation when AI is unavailable. Each writes its participation into the decision record under the same authority structure that governs human contribution.
The distinction matters: most AI assistance reaches a decision from outside it, shaping the reasoning without being subject to the institution's governance. In MagnaRix, the assistance joins the decision itself, bounded by the authority of the decision, attributed in the record, and reviewable later by any participant with access to revisit it. When AI is configured, the aids augment their core capability with language-model assistance; when it is not, they continue to produce useful, attributable results within their bounded scope.
Grounded retrieval, traceable to its source
Ask MagnaRix is a question interface that retrieves across the decision record: every decision, every review note, every comment, every change. Answers are grounded in the record, and the citations that support them are not text references. They are structured links to the specific elements being cited: the parts of the decision, the moments when changes were made, the review notes that scrutinized them. In the user interface, every citation is clickable and navigates to the cited evidence.
Each answer carries an indicator of how strongly it is grounded, derived from the breadth and depth of the evidence it draws on. Outputs with weak grounding are marked as such. The user sees the chain from question to answer to evidence, and the basis on which an answer should be trusted.
Inquiry within the decision itself
Copilot brings the same grounded retrieval into the decision itself. Within any decision, the user can ask: How was this approached? What precedent shaped it? What risks were weighed? Who authorized it? What did it change? What does it propagate into? Where does it conflict with prior reasoning?
Every answer is grounded in the decision record. Every Copilot interaction is preserved with the question, the type of inquiry it represented, and the strength of its grounding. The history is reviewable; the participation is auditable.
Purpose-built AI participation in the formation of a decision
Beyond retrieval, MagnaRix offers purpose-built reasoning aids that participate in the formation of a decision: examining risks not yet articulated, scrutinizing assumptions that may be fragile, assessing governance readiness, identifying which other decisions a commitment would affect, surfacing precedent, and evaluating alternatives.
Each aid is bounded by institution policy and operates within the authority structure of the decision. Each remains attributed in the record. None operates outside the governance that holds the decision itself. The full description of how this works in practice is in Product.
Assistance without displacement
Advisory Intelligence does not remove judgment from the decision. Responsibility, approval, and accountability remain with the people and governance structures that authorize important choices.
AI-assisted analysis can surface considerations a reviewer may not have reached and bring relevant precedent into view, yet authority remains explicit within the decision record. Findings from the aids enter the decision only when a reviewer with the appropriate authority accepts them; this keeps AI contribution useful without blurring the line between assistance and authorization.
When AI analysis becomes part of the auditable record
Advisory Intelligence lets institutions retain not only their decisions, but the visible AI assistance that participated in forming them. Later reviewers can examine how context was surfaced, how assumptions were made visible, and how structured analysis contributed to the decision.
The result is AI contribution that remains governed, traceable, and continuous with the rest of the record, so the institution can show not just what was decided but how the analysis behind it was produced and reviewed.
Advisory Intelligence keeps consequential decisions connected to the AI analysis that shaped them, without obscuring where authority and accountability remain.