Enterprise WOTC Processing Suite

$200K Revenue.
1 Week. 1 Developer.

How a solo developer replaced a $5,000/applicant phone bottleneck with a $0.10 self-service system — and generated $200K+ in year one.

$200K+
Revenue Generated
1 Week
Build Time (Solo)
50,000×
Cost Reduction
7
Languages

The $5,000 Question

IRS Form 8850, Question 13: “Were you unemployed, looking for more paid work, or felt your skills were underutilized?”

Most people qualify. But 30% answered incorrectly because they didn't understand the legal terminology. Each wrong answer = $5,000 in lost tax credits.

The Insight

Applicants weren't clicking “No” because they weren't eligible — they were clicking “No” because they didn't understand the question.

The bottleneck wasn't the code. It was human comprehension. Audio guidance solved what better UI couldn't.

Before → After

Every metric, transformed

Cost per Applicant
$5,000$0.10
Verification Time
45 minutes60 seconds
Error Rate
30%<5%
Accuracy
70%95%+

The Pipeline

End-to-end automation from email to federal compliance

Gmail API Watch → Email Processing → AI Extraction → PostGIS Queries → Dashboard → CSV Export
       ↓                  ↓                 ↓                ↓               ↓            ↓
 Auto-fetch         Parse PDFs        86 fields         EZ/TEZ zones      Kanban      Federal
   emails          attachments        extracted          verified          view       compliance
86
Fields Extracted per PDF
<50ms
EZ Zone Detection
10,000+
Geospatial Polygons

Technical Stack

Production-grade. Every layer deliberate.

Frontend
Next.js 14, React 18, TypeScript, Tailwind CSS, shadcn/ui, TanStack Table
Backend
Supabase, PostgreSQL 15, PostGIS, Row-Level Security, Edge Functions
AI/ML
OpenAI GPT-4 extraction, Gemini 2.5 classification, Claude field validation
Geospatial
PostGIS spatial queries, OpenCage geocoding, 10,000+ EZ/TEZ polygons
Integrations
Gmail API, Google Cloud Storage, Railway (5 microservices)
Infrastructure
Vercel (7 deployments), Railway webhooks, Supabase real-time
i18n
i18next — English, Spanish, French, Haitian Creole, Korean, Russian, Chinese

Geographic Intelligence

PostGIS spatial queries for Empowerment Zone eligibility

-- Automated EZ detection: database trigger → geocode → spatial intersection
SELECT COUNT(*) > 0 as is_ez
FROM empowerment_zones ez
WHERE ST_Contains(
  ez.geometry,
  ST_SetSRID(ST_MakePoint(longitude, latitude), 4326)
);

-- Result: <50ms per query across 10,000+ polygons, 3,143 U.S. counties
-- Each EZ match = $3,000 bonus tax credit automatically captured

Human Verification Layer

AI compresses to options. Humans decide.

3s
Auto-advance timer
Defaults to AI extraction
1
Click to override
Fast human correction
100%
Audit trail
Every decision logged
“The industry focused on technology — better forms, faster processing. I focused on comprehension. Audio that explains.”

Result: 50,000× cost reduction. $200K revenue in year one.
Built in Beit Shemesh, Israel. Solo. 1 week.