Schenk & Associates
1 of 12
Internal Overview

Autonomous Emerging
Tort Scanner

A system that scans regulatory, scientific, legal, and social sources every night and surfaces potential mass tort opportunities for our review each morning.

January 2026

The Idea

What is this?

  • An automated system that runs every night while we sleep Scans 15+ sources: FDA, CPSC, PubMed, legal news, Reddit, investigative journalism
  • Uses AI to evaluate what it finds against our criteria Scores each item on harm, causation, defendant viability, legal theory, and market gap
  • Researches the best opportunities in depth Produces a research memo, competitive analysis, and draft landing page copy
  • Delivers everything to a shared Google Drive folder by morning We review over coffee and decide what to act on
Why It Matters

The advantage is speed

  • Most firms find out about emerging torts months after the first signals appear They wait for MDL announcements, competitor ads, or legal news coverage
  • This system catches signals within 24 hours of publication FDA recalls, new PubMed studies, CPSC complaints, Reddit discussions
  • Earlier discovery means lower-cost client acquisition and better positioning Before the space gets crowded with competing firms and expensive advertising
  • It also monitors sources most firms never look at Scientific literature, corporate SEC filings, patient forums, investigative journalism
How It Works

Five stages, fully automated

01
Collect
Pull new items from 15+ sources across regulatory, medical, legal, and social channels
02
Filter
AI evaluates every item on five criteria and scores them 1 to 10
03
Research
Deep dive on the top-scoring topics: evidence, litigation, defendants, competition
04
Draft
Generate landing page copy with appropriate framing for each topic
05
Deliver
Upload everything to our shared Google Drive folder
Runs automatically every night at 3:00 AM Pacific. No manual input needed. Results are waiting in Google Drive by morning.
Sources

What it monitors

Regulatory & Government

FDA adverse events, drug safety alerts, recalls. CPSC product recalls. NHTSA vehicle defect investigations. FTC enforcement actions.

Medical & Scientific Research

PubMed studies linking products to health outcomes. Medical journal case reports. Environmental health research.

Legal News & Court Filings

JPML MDL transfer decisions. Reuters Legal. JD Supra mass tort and product liability articles.

Investigative Journalism

ProPublica investigations. The Markup (tech accountability). Reuters and Bloomberg investigative series.

Corporate Signals

SEC filings mentioning litigation reserves. Patent filings for "safer" product versions. Quiet product reformulations.

Social & Community Signals

Reddit communities (legal advice, health, parenting). Patient advocacy forums. Consumer complaint patterns.

Filtering

How it decides what matters

Every item is scored by AI against five weighted criteria. Only items scoring above a 5/10 move forward.

Harm
25%
Is there a concrete, documentable injury?
Causation
25%
Is there scientific evidence of a link?
Market Gap
20%
Are other firms already on this?
Defendant
15%
Is there a solvent, identifiable target?
Legal Theory
15%
Is there a viable cause of action?
Set to aggressive. The system flags speculative opportunities where research suggests harm, even if no lawsuits have been filed yet. We can dial this back.
Research

What the deep dive covers

For each topic that passes the filter, the AI investigates six areas and produces a research memo.

Scientific & Medical Evidence

Published studies, adverse event data, regulatory findings that support the alleged harm.

Existing Litigation

Any filed lawsuits, MDLs, class actions, or regulatory enforcement actions and their status.

Defendant Analysis

Company solvency, market position, litigation reserves, product reformulations.

Plaintiff Population

Estimated number of affected people, demographics, how they could be reached.

Competitive Landscape

Which PI firms already have pages on this, how strong their SEO presence is.

Legal Theory

Strongest cause of action, weaknesses, relevant precedents, statute of limitations.

Output

What we get each morning

Tort Scanner Output/ └── 2026-01-31/ ├── daily-summary.md ├── 01-ai-chatbot-psychosis/ │ ├── research-memo.md │ ├── landing-page-draft.md │ ├── sources.md │ └── competitive-analysis.md ├── 02-pfas-dental-floss/ │ └── ... └── rejected-items.md
Daily Summary
  • How many sources were scanned
  • How many new items were found
  • Each opportunity with its score and a one-line summary
  • Recommendation: pursue, monitor, or pass
Per-Topic Package
  • Full research memo with source links
  • Draft landing page ready for review
  • Competitive analysis
  • Consolidated bibliography
Landing pages use three framing levels: "Established" for active litigation, "Emerging" for early litigation, and "Under Investigation" for pre-litigation topics.
Examples

The kinds of things it finds

Technology Harms

  • AI chatbot-induced psychosis and self-harm
  • Algorithm-driven eating disorders in teens
  • Addictive app design patterns
  • AI hiring tools with discriminatory outputs

Chemical Exposure

  • PFAS in dental floss, cosmetics, cookware
  • Microplastics in food packaging
  • Emerging drug side effects
  • Contaminated supplements

Product Liability

  • Medical device failure clusters
  • Children's product safety defects
  • Noise-induced hearing loss from earbuds
  • Smart device data enabling abuse

Emerging Legal Theories

  • Wrongful death from AI companion apps
  • Platform negligence for harm to minors
  • Failure to warn for addictive design
  • Environmental exposure from facilities
Cost

What it costs to run

Research Associate
$70K
PER YEAR
Manual research. Limited sources. 40 hrs/week. Can monitor a handful of topics.
Marketing Analyst
$60K
PER YEAR
Tracks competitor activity. Manual keyword monitoring. No scientific literature.
Tort Scanner
~$125
PER MONTH
15+ sources. AI analysis. Runs 365 nights/year. No staff. Scales without added cost.
Breakdown: ~$7/month for cloud hosting (Railway), $2-5/day for AI processing (Claude API). No staff time required beyond reviewing the morning output.
Status

Where we are

Phase 1
Foundation COMPLETE
Project structure, configuration, source definitions, Google Drive integration code.
Phase 2
Collection & Filtering
Connect to live data sources (FDA, CPSC, Reddit, PubMed). Build the AI evaluation pipeline.
Phase 3
Research & Drafting
Deep research module. Landing page generation. Full end-to-end pipeline.
Phase 4
Deployment
Deploy to cloud. Set up nightly schedule. First autonomous runs.
Ongoing
Tune & Expand
Add more sources. Adjust sensitivity based on output quality. Expand coverage areas.
Next Steps

What we need to move forward

  1. Green light to proceed to Phase 2
  2. Set up a shared Google Drive folder for scanner output
  3. Begin testing with live data sources
  4. First full autonomous run targeted within the month
  5. Review first week of output together and decide on adjustments