MEDICAL DEVICE QUALITY & REGULATORY TEAMS

Keep requirements connected to their evidence.

Review document changes, follow their impact and bring missing evidence into view. Tracewren is being developed for teams working with MDR technical documentation and ISO 13485 quality records.

Early-stage project · Interactive sample available

ILLUSTRATIVE PROJECTREV 02

A review that starts with the source

REQ-014Requirement revised

The interface shall display an alert within 1 second.

RRisk controlRISK-007 · v3Review needed
DDesign recordDES-021 · v2Review needed
VVerification recordTEST-008 · v1New evidence missing

A recorded pass against revision 1 does not verify revision 2.

Review this example
MDR documentation workflowsISO 13485 document reviewEngineering traceability

THE PRODUCT

Four practical parts of a document review.

The sample workspace demonstrates the review flow. AI-assisted analysis and customer document processing are planned capabilities.

01

Traceability review

Follow requirements through risk, design and verification records. Keep proposed links separate from confirmed relationships.

02

Evidence review

Locate supporting documents and their revisions. Make absent or outdated evidence visible before a review meeting.

03

Change impact review

Compare revisions and identify connected records that may need reassessment. Each finding stays attached to its source.

04

Review actions

Record an owner, a target date and a review decision. Export the sample findings as a structured review record.

WHERE IT FITS

Built around the records your team already works with.

MDR technical documentation

Organize the evidence supporting your technical-file review. Use a team-defined checklist and keep the document behind each finding visible.

ISO 13485 quality records

Review changes to procedures, design inputs and supporting records. Carry unresolved findings into a deliberate human review.

Design and verification

Track a requirement change across its linked design, risk control and test evidence. See what was verified, against which revision.

HOW IT WORKS

Source. Connect. Review.

01

Start with a version

Identify each document and revision before comparing its content.

02

Follow the links

Inspect connected records and proposed changes with their source references.

03

Make the decision

A reviewer assesses the findings. Software suggestions do not establish approval or compliance.

ONE CHANGE, THREE QUESTIONS

A faster alert needs more than an updated sentence.

In our fictional example, an alert requirement changes from two seconds to one. The risk control needs review, the design still cites the old requirement, and the earlier test has not been rerun.

Review this example
REQ-014 · REV 012 seconds
REQ-014 · REV 021 second

A recorded pass against revision 1 does not verify revision 2.

ABOUT TRACEWREN

Clear evidence. Deliberate decisions.

Tracewren is an early-stage product project focused on document review and design traceability for medical device teams. We are starting with the relationship between a requirement, the records that implement it and the evidence used to verify it.

Our approach keeps missing information visible and preserves the distinction between a software suggestion and a human decision. The public demonstration uses fictional records; a production AI service is not yet available.

Questions before you start.

Can I use Tracewren with my own documents today?

The public workspace currently uses a fixed fictional example. Document upload, customer workspaces and AI analysis are not available in this demonstration. Contact us to discuss the intended workflow.

Does Tracewren certify MDR or ISO 13485 compliance?

No. The proposed product supports documentation review and evidence organization. Applicable requirements, conformity assessment and approval decisions remain the responsibility of qualified people and the relevant organizations.

What does a review acknowledgment mean?

It means a finding has been read and acknowledged in the sample. It does not mean a requirement is satisfied, a test passed or a document has been approved.

How is Claude intended to be used?

We plan to evaluate Haiku for classification and structured extraction, and Sonnet for contextual analysis. Source-reference accuracy, missed changes, latency and cost will guide model selection.

Where does the demo send my data?

The sample runs in your browser with fictional records. Its review controls do not send records to an AI service. The export is a local JSON download. Hosting providers may process normal connection logs.

What does it cost?

The public sample is free to explore. Commercial pricing has not been set. Pilot scope, availability and pricing will be agreed before any paid work.

Contact

Tell us where your evidence gets disconnected.

Have a document-review workflow to discuss? Send a short description of your team and the problem. Please do not send patient data or confidential technical files in an initial inquiry.

Our company email is being set up. Contact details will be available once the domain is active.