Skip to content
MDARIXProduct intelligence
Platform · Investigations

Investigate with the evidence, the explanations and the gaps side by side.

One governed case for each question: the evidence, competing hypotheses and what is still unknown stay visible, so a plausible story is never mistaken for proof.

  • Available today5 capabilities
  • In a pilot1 capability
Ask MDARIXAvailable today

Ask a precise question and get a cited, bounded answer.

Ask a focused question about an authorized investigation and receive a bounded response that keeps source evidence, temporal scope, uncertainty, and human responsibility visible.

AI is off during the trial: Ask MDARIX answers from your records by its rules and says so. Answers written by AI come in a pilot, once AI is validated and switched on.

  • Start with a precise question. Frame the issue, product, product version, time window, and investigation scope before retrieval begins.
  • Retrieve only what you can access. Tenant, role, field, and record policies remain authoritative; a prompt cannot widen access or reveal hidden data.
  • Separate answer from evidence. Show supporting sources, contradictions, unknowns, missing evidence, and limitations alongside the interpretation.
  • Abstain when the record is weak. Unsupported causal claims are bounded or declined rather than presented as facts.
  • Keep people accountable. AI assists investigation; authorized reviewers decide what is accepted, escalated, or sent for further evidence.
  1. Question
  2. Scope
  3. Retrieve
  4. Cite
  5. Challenge
  6. Review
Investigation WorkspaceAvailable today

Run the investigation in one governed workspace.

Move from a complaint or field signal to a structured, evidence-grounded investigation. MDARIX keeps product context, competing explanations, unknowns, provenance, and human review visible at every step.

Every case has one page: scope, evidence, analysis, reviews, the decision brief and the decision.

  • Frame the investigation. Define the question, product and product version scope, time window, owner, and review objective before analysis begins.
  • Build the evidence chain. Connect complaints, changes, components, suppliers, lots, observations, and source records without rewriting authoritative systems.
  • Compare explanations. Keep hypotheses, supporting evidence, contradictions, failure chains, and unknowns separate so a plausible story is not mistaken for proof.
  • Make gaps actionable. Show missing tests, unavailable sources, stale records, and unresolved relationships so reviewers know what to investigate next.
  • Prepare a governed decision. Assemble a versioned brief with provenance and limitations; authorized humans remain responsible for conclusions and approvals.
  1. Signal
  2. Scope
  3. Context
  4. Evidence
  5. Hypotheses
  6. Unknowns
  7. Review
Evidence IntelligenceAvailable today

Bring scattered evidence into one authorized context.

Bring quality, product, supplier, manufacturing, and post-market records into one authorized evidence context. MDARIX preserves provenance, relationships, contradictions, and gaps so teams can review before they decide.

Evidence comes from the exports you import. Live connections to your systems are built in a pilot.

  • Collect the evidence. Connect source records from QMS, PLM, ERP, supplier, service, and post-market systems while keeping each system authoritative.
  • Anchor every record. Relate observations to the right product, product version, component, supplier, lot, complaint, investigation, and time window.
  • Separate fact from inference. Keep observed information distinct from interpretations, hypotheses, recommendations, and human decisions.
  • Expose contradictions. Make conflicting values, stale records, unavailable sources, and incomplete relationships visible instead of hiding uncertainty.
  • Prepare traceable review. Carry citations, source identifiers, limitations, and reviewer responsibility into the investigation brief and decision package.
  1. Sources
  2. Context
  3. Provenance
  4. Evidence
  5. Gaps
  6. Review
  7. Decision
Hypothesis IntelligenceAvailable today

Keep competing explanations side by side.

MDARIX organizes possible explanations for a signal as separate hypotheses, each with the evidence that supports it, the evidence that contradicts it, its assumptions and its gaps, so a plausible story is never mistaken for proof.

Hypotheses, contradictions and open questions are written and judged by hand in a case's Intelligence tab.

  • More than one explanation. Capture each candidate explanation on its own instead of letting the first plausible story become the default.
  • Evidence on both sides. List supporting, contradicting and contextual evidence for every hypothesis, each linked to its source record.
  • Assumptions in the open. Write down what each hypothesis assumes, so reviewers can test the assumption rather than the conclusion.
  • Status, not verdict. Hypotheses carry a review status. MDARIX does not declare a root cause; authorized people decide what is accepted.
  1. Signal
  2. Hypotheses
  3. Support
  4. Contradict
  5. Gaps
  6. Review
AI ChallengerIn a pilot

Challenge the explanation before it becomes a conclusion.

Compare competing explanations against authorized medical-device evidence while keeping contradictions, unknowns, limitations, and human responsibility visible.

AI is off during the trial. The AI challenge of a hypothesis is switched on in a pilot once it is validated.

  • Challenge the favorite. Test the leading explanation against contradicting records and alternative explanations before it hardens into a conclusion.
  • Name the weak assumption. Point to the assumption a hypothesis depends on most, and what evidence would confirm or break it.
  • Visible to reviewers. Challenges stay attached to the hypothesis, so approvers see what was questioned and how it was answered.
  1. Hypothesis
  2. Challenge
  3. Contradictions
  4. Alternatives
  5. Review
Unknowns RadarAvailable today

See what the investigation still cannot establish.

The Unknowns Radar lists what is missing before anyone decides: incomplete traceability, missing lot references, unavailable sources, stale records and unsupported links, each with the evidence that would close it.

A case's open questions and missing evidence, written by hand, stay visible up to the decision.

  • Missing evidence, named. Show which records, tests or source systems the investigation has not yet seen, instead of silently ignoring them.
  • Broken traceability. Flag complaints without lot references, components without supplier links and changes without effective dates.
  • What would close the gap. Pair each unknown with the evidence or request that would resolve it, so the next step is obvious.
  • Carried into the decision. Open unknowns stay visible in the decision context, so nobody signs off without seeing what is still uncertain.
  1. Scope
  2. Expected evidence
  3. Found
  4. Missing
  5. Request
  6. Review
Availability

Every capability says where it stands.

The labels are the roadmap's: nothing is marked Available today until it works in a free trial workspace.

Available today

In every workspace, including the 30-day free trial.

In a pilot

Built and validated together during a pilot.

Planned

Not available yet.

See the roadmap
Design partners

Bring one real investigation. See the whole story in one place.

Start with exports from your quality, product and supplier systems. Your systems of record stay where they are.

Start free trial