Product descriptions can only tell you so much. What matters is what the workflow actually looks like — what the system does, in what order, and what happens at each step. LuminaNarrate is a fully automated pipeline. There is no manual review step inside the platform. Here is the complete flow from file upload to delivered output.
LuminaNarrate handles everything from intake to delivery automatically. The AI both generates and reviews the output. Your company's reviewers receive the finished document by email — they never need to log in to the platform to do their review.
Step 1 — PDF Upload
The workflow begins when a source document is uploaded. LuminaNarrate accepts adverse event case data in PDF format — source narratives, case report forms, or raw documentation from any origin. Files are uploaded through the platform's intake interface, which validates format, file size, and completeness before processing begins.
Document Intake
Upload the source PDF through the secure file intake interface. The system accepts up to 20 files per batch at 50 MB max per file. Once uploaded, the document enters the automated processing pipeline immediately — no configuration required per case.
Step 2 — MedDRA Coding & WHO Drug Coding
With the source document parsed, the pipeline runs two coding processes in parallel. MedDRA coding maps reported adverse events to the current MedDRA hierarchy — system organ class, high-level group term, high-level term, and preferred term. WHO Drug coding identifies and standardises suspect and concomitant medicinal products against the WHO Drug Dictionary.
Automated Dual Coding
Both MedDRA and WHO Drug coding run automatically with no human input required. The AI applies the applicable coding standard version configured for your account. Coding results are embedded directly into the generated output documents and are visible to your reviewers alongside the narrative.
Step 3 — Narrative Generation
Using the structured data extracted from the source document and the completed coding results, LuminaNarrate generates a complete case narrative. The narrative follows the ICH E2B format: patient details, suspect product(s), reported events, relevant medical history, course of events, and outcome — all derived from the uploaded source data.
Full Narrative in Seconds
The AI constructs the narrative directly from the source document data. No templates to fill in, no drafting from scratch. The output is a complete, structured narrative with MedDRA and WHO Drug codes embedded, ready for the AI review stage that follows.
Step 4 — AI Review, Flags & Confidence Score
Once the narrative is generated, a second AI layer reviews the output. This is not a human review step — it is a fully automated quality pass. The review AI examines the narrative for consistency, completeness, coding alignment, and potential anomalies. It then makes one of two determinations: the case can proceed automatically, or it should be flagged for attention by the company's reviewers.
AI Quality Pass with Confidence Scoring
Every processed case receives a confidence score embedded in the output document. Where the AI review identifies a discrepancy, a data gap, or a coding ambiguity, it raises a flag directly in the output — clearly marked so the receiving reviewer knows exactly what to examine. Cases with high confidence and no flags proceed to delivery automatically. Flagged cases are delivered with the flags intact so reviewers have full context.
There is no manual review inside LuminaNarrate. The platform does not hold cases in a queue waiting for an in-app approver. All review happens on the output document — by your company's authorised reviewers — outside the platform, after delivery.
Step 5 — Automated Delivery to Authorised Emails
When processing is complete, LuminaNarrate automatically sends the output documents to the authorised email addresses configured for the customer account. These addresses are set up at account creation and are the only destinations the system will ever send output to — no ad-hoc recipients, no manual routing decisions.
Automatic Email Delivery
Output — including the narrative, MedDRA and WHO Drug coding, AI confidence score, and any flags — is delivered automatically to your configured authorised recipients the moment processing and AI review complete. No one inside the platform needs to trigger the send. If delivery needs to be repeated, a Resend option is available directly from the case record.
What Happens When a Case Fails
Occasionally a case may fail during processing — due to a malformed source document, an unexpected data structure, or a transient pipeline issue. LuminaNarrate does not silently discard failed cases. The failure is surfaced in the dashboard with a clear status, and a Reprocess option is available immediately. Reprocessing re-runs the full pipeline from the point of failure with no data loss.
Reprocess on Failure
Failed cases appear in the dashboard with a distinct status indicator. The Reprocess action re-runs the pipeline automatically. Once reprocessing completes successfully, the output is delivered to authorised recipients in the same way as any other completed case — no manual intervention required beyond triggering the reprocess.
The Complete Automated Loop
The full pipeline — upload → MedDRA & WHO Drug coding → narrative generation → AI review with confidence scoring and flags → automatic email delivery to authorised recipients — runs without requiring any human action inside the platform. Your team interacts with the output documents they receive. LuminaNarrate handles everything up to that point.
For teams processing large volumes of adverse event cases, this means consistent throughput without the bottlenecks that come from in-platform manual review queues. Every case is handled the same way, every time.