Five specialized AI agents debate your dilemma from opposing perspectives, then synthesize a structured recommendation with quantified risk and confidence metrics.
Most AI tools give you one answer from one perspective. Real decisions are messier than that — they involve tradeoffs between risk, growth, values, and timing that a single viewpoint misses. VerdictAI X runs your decision through five adversarial agents who challenge each other, then distills the debate into a final verdict.
1. Input Ingestion The system parses your decision, stated constraints, and implicit goals to establish a baseline analytical state.
2. Independent Analysis Five agents analyze the situation concurrently, each from a distinct professional and psychological lens.
3. Adversarial Debate Two rounds of live debate. Round one: agents attack vulnerabilities in each other's reasoning. Round two: agents defend or revise their position based on the pushback received.
4. Final Synthesis A dedicated Verdict Engine resolves conflicts, extracts consensus, and produces a final recommendation with quantified risk, growth potential, and confidence scores.
| Agent | Role |
|---|---|
| The Strategist | Optimizes for long-term ROI, mathematical leverage, and compounding returns |
| The Guardian | Risk-mitigation layer — calculates financial runways and stress-tests assumptions |
| The Visionary | Scans for non-linear upside and exponential opportunities that others overlook |
| The Humanist | Evaluates psychological cost, relationship impact, and alignment with personal values |
| The Contrarian | Interrogates the framing itself — surfaces hidden biases, shortcuts, and systemic flaws |
| Layer | Technology |
|---|---|
| Analysis LLM | Gemini 1.5 Flash — high-speed concurrent agent reasoning |
| Synthesis LLM | Gemini 1.5 Pro — high-context final verdict generation |
| Interface | Gradio with custom dark-themed CSS layer |
| Backend | Python with async streaming for real-time output |
| Deployment | Hugging Face Spaces |
| Security | Environment-based secret management, no keys in source |
git clone https://github.com/A-Square8/VerdictAI-X
cd VerdictAI-X
pip install -r requirements.txt
cp .env.example .env
# Add your Gemini API key to .env
# Get a free key at Google AI Studio
python app.pyBuilt as a technical demonstration of multi-agent orchestration and adversarial reasoning pipelines.