In June 2025, I couldn't write a line of code. I'd never opened a terminal. The closest I'd come to software development was building spreadsheets for my medical distribution business.
Twelve months later, I have a market intelligence engine running 34 autonomous signal channels, scoring 157 tickers daily, with zero human input. It produces a morning brief before I wake up and a closing analysis before I finish dinner.
I didn't hire engineers. I didn't raise venture capital. I didn't go through an accelerator. I built it with AI coding agents — and 16 years of knowing exactly what I wanted to build.
This is the story of how DEFINTEL went from a blank folder on my desktop to a production system that runs autonomously every market day.
The Background Nobody Asks About
Before I touch the technology, you need to understand where the ideas came from.
I ran a medical distribution business for 16 years. Not a startup — an actual operation. Customer service, order fulfillment, shipping, inventory management, regulatory compliance, accounting. All of it. By myself.
DSCSA compliance alone — the FDA's Drug Supply Chain Security Act — required tracking every unit of product from manufacturer to patient. Serialization. Verification. Transaction documentation. I spent years doing this manually before building systems to automate it.
That experience taught me something that most software founders don't understand: the majority of what businesses spend hours doing every day can be automated. Not everything — you can't automate filling orders and putting them on a truck. But the analysis, the tracking, the reporting, the pattern recognition? That's software's job.
I also played poker professionally for seven years. That's relevant because poker is fundamentally about processing incomplete information under pressure. You never have all the data. You're making probabilistic decisions with partial knowledge, reading patterns in behavior, and managing risk with every action.
Trading is the same problem with different inputs.
The First Month: June 2025
I started with Claude — Anthropic's AI assistant — and a simple question: "Can you help me build software?"
The answer was yes, but not in the way most people think about AI-assisted coding. I wasn't asking Claude to write a script and then trying to understand what it did. I was describing business problems I'd spent 16 years solving manually, and having Claude translate my operational knowledge into code.
The first project wasn't DEFINTEL. It was FORGE — an ERP system for my distribution business. I knew exactly what the software needed to do because I'd been doing it by hand for over a decade. Claude wrote the code. I validated that it solved the actual problem.
Within weeks, I realized something: the bottleneck wasn't the coding. It was the conversation. One Claude instance at a time was too slow. I needed parallelism.
The 20-Clone Method
Here's what nobody tells you about AI-assisted development: you can run multiple instances simultaneously.
I set up 20 copies of my codebase using git worktrees — each one on its own branch, each one with its own Claude instance. I'd write detailed prompts for each instance, deploy all 20 at once, and merge the completed work.
My development machine (24-core Intel Ultra 9 285K, 32GB RAM, RTX 5080) could handle it. What I called "the swarm" became my development methodology.
On a good day, 20 Claude instances working in parallel could produce what a small engineering team might ship in a week. The key was preparation — each instance needed a clear scope, a well-defined interface, and enough context to work independently.
This is how the codebase reached 90,000+ files in 12 months. Not by working faster, but by working wider.
Why DEFINTEL Exists
After building FORGE (ERP), JADE (AI website builder), and PIPER (sports betting intelligence), I turned to the market I understood most deeply: trading.
The problem with retail trading intelligence is fragmentation. You check TradingView for technicals. You check Unusual Whales for options flow. You check Quiver Quantitative for congressional trades. You check another site for dark pool data. You check another for gamma exposure.
Each source gives you one slice. Nobody gives you the composite picture.
I wanted a system that:
- Ingested data from every meaningful source
- Scored each ticker across all channels simultaneously
- Identified patterns a human would miss (correlations between congressional trades and dark pool activity, for example)
- Ran autonomously — no morning routine of checking 8 websites
DEFINTEL is that system.
The 34 Channels
Each channel is an independent signal processor. They don't communicate with each other during analysis — each one evaluates every ticker on its own terms, and the composite scoring engine weighs their outputs.
Dark pool block trades (FINRA ATS data), unusual options activity, volume profile analysis, order flow imbalance
13F filings, insider transactions (Form 4), institutional ownership changes, short interest (bi-weekly official + daily estimates)
Congressional SEC filings (EFDS), committee hearing correlation, FDA calendar events, regulatory action tracking
Gamma exposure mapping, dealer positioning models, put/call ratio analysis, volatility term structure, skew analysis
VIX term structure, credit spread monitoring, market breadth analysis, sector rotation tracking, correlation regime shifts
Prediction markets (Polymarket, Kalshi), AIS shipping data, weather events, social sentiment velocity, news NLP
Each channel has its own data source, its own refresh cycle, and its own scoring methodology. The composite engine weights them differently per ticker — a pharmaceutical stock weights FDA channel signals higher than a technology stock would.
What the System Produces
Every market day, DEFINTEL outputs:
- Morning brief — Pre-market analysis: regime state, top-scoring tickers, overnight developments, key levels to watch
- Signal heatmap — All tracked tickers with composite scores, sortable by score, sector, or individual channel
- Congressional trade tracker — Every SEC filing from members of Congress, AI-scored for significance
- Contagion map — When one ticker moves, which others historically follow within 1–5 days
- Gamma cliff alerts — Which tickers are approaching dealer gamma flip levels
- Closing brief — End-of-day summary: what the signals caught, what they missed, regime changes
The system runs without me. I check the brief in the morning the same way I'd check the weather.
The Part I Can't Believe Worked
The most technically ambitious piece is the contagion mapping. When NVDA makes a significant move, DEFINTEL traces historical correlation patterns to predict which tickers tend to follow within 1–5 trading days.
This isn't simple correlation — it's directional, time-lagged, and sector-aware. A move in NVDA affects AVGO and AMD differently than it affects MSFT or AAPL. The system builds these correlation chains from historical data and updates them weekly.
When I first saw the contagion map light up after a major NVDA earnings move and correctly highlight that AVGO would follow within 48 hours, I genuinely couldn't believe I'd built it. Not because the code was complex — Claude wrote that. But because the analytical framework that told Claude what to build came from years of watching these patterns manually and never having the tools to systematize them.
That's the part people miss about AI-assisted development. The AI doesn't have the domain knowledge. It has the coding ability. The operator — the person with 16 years of business experience and 7 years of pattern recognition from poker — provides the domain knowledge. The AI provides the implementation.
What It Costs
Nothing, if you want the free tier. Congressional trades, regime state, top signal scores, SPY gamma, and a morning brief. Permanently free. No trial. No credit card.
Starter ($14.99/mo), Intelligence ($59.99/mo), and Professional ($149.99/mo, limited-time offer) unlock the full system — real-time data, all channels, interactive contagion mapping, dark pool access, squeeze detection, and more.
For context, Trade Ideas charges $228/mo for their premium tier. Bloomberg Terminal is $24,000/year. I'm not comparing DEFINTEL to Bloomberg — different scale, different audience. But for independent traders who want more than a price chart and a Reddit thread? This is built for you.
What I Learned
1. Domain expertise beats coding skill. I can't write a Python decorator from memory. But I can describe exactly what a contagion mapping system should do, what data it needs, how the output should look, and what edge cases to handle. That's more valuable than syntax knowledge.
2. AI development is parallel, not serial. The 20-clone method isn't a hack — it's a paradigm. Traditional development is bottlenecked by human cognitive bandwidth. AI-assisted development is bottlenecked by how well you can decompose problems and define interfaces. Those are different skills.
3. Ship the damn thing. I could have spent another 6 months refining the scoring model. Instead, I opened it for founding members with a free tier that's genuinely useful and a feedback loop that makes it better every day. The system improves autonomously. Waiting for perfection was the wrong instinct.
Try It
The free tier is live: defintel.net
The congressional trading tracker requires no signup: defintel.net/congressional
Founding members get 7 days of full Elite access: defintel.net/launch
I built this alone. I'm improving it alone (with AI). If you have feedback, questions, or signal channel suggestions, find me on Discord or email chad@justiceapex.ai.
No VC. No team. No coding background. Just AI agents and an operator who spent 16 years learning what needed to be automated.