Case Studies
Real transformations. Real results. Actually deployed. Actually running.
WOTC Tax Credit Transformation
The Context
WOTC (Work Opportunity Tax Credit) allows employers to claim $2,400–$9,600 per eligible hire. It's a federal program designed to incentivize hiring from disadvantaged groups.
The problem: Most employers leave massive money on the table.
The Problem
Question 13 on the WOTC form asks whether the employee was unemployed, wanted more paid work, or felt underemployed when they started the job.
Here's what happens:
- ×Employee sees confusing legal language
- ×Doesn't understand what it's actually asking
- ×Clicks “No” to be safe
- ×But they actually qualify
- →Employer loses $5,000+ in tax credits
Multiply this by hundreds of employees. You're looking at six figures in lost tax credits annually.
The industry had been trying to solve this for years. Better form design. Clearer instructions. Training programs. Nothing worked. Wrong answers remained stubbornly high: 30%.
The Insight
“Applicants weren't clicking 'No' because they weren't eligible. They were clicking 'No' because they didn't understand the question.”
This wasn't a UI problem. It wasn't a training problem. It was a comprehension problem.
The bottleneck wasn't technical—it was human understanding.
The Solution
I built an audio-guided form.
Every question is read aloud in plain language. The employee hears what the question actually means. They answer correctly. The tax credit is captured.
The Results
| Metric | Legacy Methods | Audio Form |
|---|---|---|
| Wrong answers | 30% | < 5% |
| Credits LOST | $5,000+ each | Nearly zero |
| Completion time | 45 minutes | 60 seconds |
| Build time | — | 1 week |
| Credits CAPTURED | Many missed | $200K+ |
One person. One week. $200K in tax credits captured that legacy methods missed.
The insight wasn't “build a better app.” The insight was “the bottleneck is comprehension, not code.”
Timeline
The System Components
Audio WOTC Verification
$200K+ credits capturedVoice-guided form that reads questions aloud in plain language. 60 seconds to complete.
Digital IRS Form 8850
100% form coverage7 languages. Touch signatures. Real-time validation. Mobile-friendly.
WOTCFY Platform
Enterprise-grade29K LOC. AI form extraction. PostGIS EZ zone detection. 4-stage verification pipeline.
Brain MCP — Cognitive Prosthetic
The Context
Asperger's + ADHD = monotropic attention. One tunnel at a time. Total immersion — and total amnesia for everything outside it.
Every context switch was a factory reset. Decisions remade. Questions re-asked. Breakthroughs re-discovered months later.
The Problem
Context dies when the tunnel moves. Three months deep in Torah study? The AI architecture decisions don't exist anymore. Switch to frontend? The 47 open Torah questions vanish. Not deprioritized. Gone.
This isn't poor organization. It's neurology. Monotropic attention doesn't do “background threads.”
The Insight
“The Bottleneck IS the Amplifier. Don't fix monotropic attention — build infrastructure that preserves context across attention shifts.”
Log everything. Embed it. Make it queryable. Let the tunnel do what it does best. Give it a safety net.
The Solution: 25 MCP Tools
377K messages indexed across Claude, ChatGPT, Claude Code, and Clawdbot. 82K semantic embeddings for conceptual search. 9,979 structured summaries with extracted decisions, questions, and breakthroughs.
+ 4 more prosthetic tools (open_threads, dormant_contexts, cognitive_patterns, tunnel_history) and 17 generic tools (semantic search, thinking trajectory, alignment check, etc.)
The Results
| Metric | Before | With Brain MCP |
|---|---|---|
| Past context | Gone when tunnel moves | 256ms retrieval |
| Open questions | Forgotten permanently | 111,942 preserved |
| Decisions | Remade from scratch | 36,743 tracked |
| Domain switching | Full restart | Cost quantified + brief |
| Context survival | 0% | 100% |
One person. 18 months. A cognitive prosthetic that turns neurological context death into queryable institutional memory.
The insight wasn't “fix the attention system.” The insight was “the bottleneck IS the amplifier — build a safety net around it.”
Timeline
Technical Stack
More Case Studies
YouTube Pipeline
32K videos transcribed. 41.8M words processed. Local ML, zero cost.
Google Apps Script Portfolio
41 production scripts. SSN matching, Torah translation, WooCommerce automation.
Have a bottleneck that needs finding?
48 hours to identify it. 5 days to prove the solution. Free until value is demonstrated.
Let's Talk