Adverse Intake

A guided medication-interaction intake demo that keeps label evidence and human review in view.

Overview

Adverse Intake turns a synthetic report of a suspected medication interaction into a structured draft for human review. The guided conversation captures the products, reported experience, and timeline while showing relevant label evidence and clinical-review signals.

I built the local demo around a deterministic intake workflow, structured validation, and curated contraindication data. The separate Proveria demonstration shows how verified software, authorized actions, and portable receipts can strengthen a regulated workflow; external trust and AI paths are optional rather than required for the local intake.

Screenshots

View full gallery

△ Problem

A suspected interaction can be difficult to document consistently, and AI-assisted workflows in regulated settings need evidence, accountability, and a clear human-review boundary.

✓ Solution

I designed a guided, synthetic intake that organizes the report, checks versioned label evidence, and prepares a draft record without making a treatment decision.

↗ Outcome

✓ Structured a review-ready synthetic intake
✓ Surfaced contraindication evidence and review signals
✓ Kept clinical decisions with qualified human reviewers

Key Features

▥

Guided Intake

Collects the reporter, products, experience, and timeline one step at a time.

◉

Label Evidence

Checks curated contraindication assertions from a versioned snapshot.

△

Review Signals

Flags medication-safety concerns for qualified human review.

⇩

Draft Export

Assembles a client-owned case record that can be downloaded as JSON.

✓

Clinical Boundary

Documents the report without making an autonomous treatment decision.

⌁

Optional Proof Path

Illustrates verified software, authorized actions, and receipts through Proveria.

Tech Stack

Next.js 16React 19TypeScriptZodVersioned JSONEvidence Workflows

↗

Make every intake review-ready.

Explore how Adverse Intake brings structured reporting, label evidence, and human review together.

View Case Study See Next Project