Adverse Event / Reportability Decision Tool

The problem this kills

A complaint comes in. Somewhere in it is a question your team has to answer correctly and document defensibly: is this reportable? Did harm occur? Could the product have caused or contributed to a serious injury? Does it meet your regulator's reporting criteria — and if it does, what's the deadline, and when did that clock actually start?

Today that decision lives in someone's head, an email thread, and a spreadsheet. The reasoning gets summarized after the fact, if at all. The deadline is tracked by memory or a calendar reminder that the wrong person owns. And the part auditors scrutinize most — the "not reportable" calls — often has the thinnest paper trail, because nobody documents a decision to do nothing as carefully as a decision to file.

That's the failure mode that hurts: a late report because the awareness-date clock was never started, or a "no" decision you can't defend two years later because the rationale was never written down. This is exactly the kind of rules-based, high-stakes, must-be-documented decision that a small internal tool handles better than a spreadsheet — and you do not need to be a developer to build it.

What you'll build

A simple internal web tool for your regulatory and quality team. You load the complaint or adverse event — what happened, the product, whether harm occurred, the dates. The tool walks you through your regulator's reportability decision tree one question at a time, and AI drafts the rationale for where each answer lands — the documented reasoning behind both the "yes" and the "no." For a reportable event it also drafts a report skeleton pre-filled with the fields your regulator requires, and starts a deadline clock counting from the awareness date.

Then the human gate: a qualified regulatory reviewer reads the determination and its rationale, edits anything that's wrong, and approves the reportable / not-reportable call. For reportable events the reviewer separately approves the report content before it's considered ready to file. Resend sends deadline reminders so nothing slips. The tool assists and documents the decision — it never decides, and it never files, on its own.

What's inside the Implementation Plan

The plan is a single file you paste into an AI coding agent. It opens by interviewing you about your business — which regulation and regulator you report to, the exact branches and criteria of your decision tree, what counts as "serious injury" and "could cause or contribute," how you define the awareness date, your reporting timelines, the report fields your regulator requires, and your nastiest edge cases — and then tailors the decision tree, the data model, the clock, and every later step to your answers. This is a tool shaped around how you make reportability calls, not a generic template.

From there it walks the agent through the database schema, loading an event, the guided decision tree with AI-drafted rationale, the reviewer's approve-the-determination gate, the report skeleton and the deadline clock from the awareness date, the reviewer's approve-the-report gate, Resend deadline reminders, and the export of the decision and report log. Every step ends with a ready-to-copy prompt. Because the whole thing runs on data you enter (or a CSV / Google Sheet import) and produces a clean CSV export in the columns your system expects, you can build and use it this weekend even with no connection to your complaint system.

The governance it includes (this is the point)

This documents regulatory decisions, so it is built like it matters: login so only your team can use it, row-level security so you only ever see your own organization's events, and a complete audit trail of every entry, edit, determination, approval, and export — who did what, and when. Nothing is treated as decided automatically: the determination is a draft with AI-drafted reasoning until a qualified reviewer approves it, and approval is the hard human-in-the-loop gate. A reportable event needs a second approval — of the report content — before it's ready to file. Duplicate guards keyed on the event ID mean the same event can't be opened, determined, or filed twice. And the tool is explicit that it assists, it does not provide regulatory or legal advice — the qualified reviewer owns the call.

Who it's for

Regulatory affairs and quality professionals in regulated industries — medical devices, food, pharma, consumer products — who must make and document reportability determinations against a regulator's criteria and deadlines. If you can explain how your decision tree works and when the clock starts, you can build this.

You've got this — open the plan, paste the first prompt, and let it interview you about how you make the call.