Grounds ✨

Transform complex decisions into clear, auditable outcomes.

🚀 Live Demo

▶️ Video Demo

Features • Quick Start • Architecture • API Reference

🎯 What is Grounds?

Grounds is a decision intelligence workspace that helps you think better, not faster.

In a world where AI can generate endless opinions, Grounds takes a different approach: it doesn't tell you what to decide—it helps you see your decision clearly.

"A mirror, not a verdict."

⚡ The Problem

Every day, professionals face high-stakes decisions across every industry:

Industry	Example Decision
👔 Business	A CEO choosing between acquisition targets or crisis response strategies
🏥 Healthcare	A hospital board deciding on AI-assisted diagnosis adoption
⚖️ Legal	A law firm choosing between AI tools and human-only positioning
💰 Finance	A wealth manager evaluating AI portfolio management amid unclear regulations
🏛️ Government	A city council weighing smart city investment vs. citizen privacy
🕊️ Religion	A pastor considering AI tools while preserving spiritual authenticity
📚 Education	A university deciding how to integrate AI tutors with academic integrity
🌍 Environment	A company choosing between AI carbon management and traditional audits
🎬 Media	A studio deciding on AI actors vs. human-only brand positioning
🚗 Transportation	A city approving autonomous vehicle zones amid safety concerns

These decisions share common challenges:

Challenge	Pain Point
😵 Information Overload	Too much data, not enough clarity
🔮 Hidden Assumptions	Blind spots you don't know you have
⏰ Time Pressure	No time for proper analysis
📝 No Audit Trail	Can't explain decisions 6 months later

💡 The Solution

Grounds provides a structured thinking environment with a clear workflow:

Step	Action	Outcome
🎨 Theme	Select your industry/domain	Contextual templates & case studies
📝 Input	Upload documents, images, or type context	Structured decision context
🔍 Gemini Critic	AI hard-edge critique with Gemini 3	Blind spots & assumption challenges
⚖️ Provider Compare	Run through multiple AI providers	Multi-perspective consensus analysis
💬 Sentiment Analysis	Aspect-based tone detection	Emotional undercurrents revealed
🧠 Deep Reasoner	Adversarial stress testing	Robustness scores & failure scenarios
📊 Export Report	Generate professional reports	Shareable HTML/PDF with full audit trail

✨ Features

🎤 Voice Input (Multimodal)

Speak your decision context using voice input:

Feature	Description
🎙️ Voice Recording	Click microphone to start dictating
📝 Live Transcription	See your words appear in real-time
🔄 Append or Replace	Add to existing text or start fresh
🌐 Browser Native	Uses Web Speech API, no external services

How it works:

🎤 Click Mic → 🗣️ Speak → 📝 Transcribe → ✅ Insert Text

📖 About & Guided Tour

Interactive onboarding for first-time users and hackathon judges:

Feature	Description
📖 About Modal	Complete guide with 5 tabs: Overview, Workflow, Reports, Benefits, Creator
🎯 Guided Tour	Step-by-step walkthrough with highlights
💡 Tooltips	Contextual help on every button
🏆 Gemini Showcase	Prominent branding for hackathon judges
🎨 Animated Logo	Vibrant gradient ring with rotating animation
👤 Creator Info	Social links (Twitter, GitHub, Discord)

About Modal Tabs:

🎯 Overview — What Grounds does and why
📋 Workflow — 7-step decision process (expandable details)
📊 Reports — HTML/PDF export features
⭐ Benefits — Key advantages for users
👨‍💻 Creator — About the developer

Tour covers: Theme → Input → Gemini Critic → Provider Compare → Sentiment Analysis → Deep Reasoner → Export Report

📤 Google Workspace Export

Export reports directly to Google ecosystem:

Destination	What's Exported
📄 Google Docs	Text summary (copy & paste)
📊 Google Sheets	Data tables as TSV (copy & paste)
📑 PDF Download	Full report with all visualizations

Integration: OAuth-based secure export (or clipboard copy for quick sharing)

🔍 Gemini Web Search Integration (Grounding)

Real-time research powered by Gemini Grounding with Google Search:

Feature	Description
🌐 Auto-Research	Toggle to automatically research your decision context
📚 Source Citations	Real URLs from Google Search results
🔑 Key Findings	AI-extracted insights from search results
⚡ Low Latency	Fast grounded responses with source verification

How it works:

🎯 Decision Context → 🔍 Gemini Grounding → 📚 Google Search → 💡 Synthesized Insights

📋 Decision Templates & Template Gallery

Quick-start templates for common decision scenarios with a searchable gallery:

Category	Example
💻 Technology	Cloud Migration, AI Tool Adoption, Infrastructure Scaling
🏥 Healthcare	EHR System Selection, AI Diagnosis Adoption
💰 Finance	Investment Decision Analysis, Portfolio Strategy
⚖️ Legal	AI Contract Analysis, Compliance Strategy
📚 Education	AI Tutors, Academic Integrity Policies
🏠 Real Estate	Property Investment, Market Analysis
✈️ Travel	Business Travel, Fleet Management
👥 HR	Remote Work Policy, Talent Acquisition
🔬 Research	Clinical Trials, R&D Investment
🚀 Startup	Funding Strategy, Pivot Decisions
🌐 General	Crisis Response, Strategic Planning

Template Gallery Features:

🔍 Search & Filter — Find templates by keyword or category
📋 One-Click Load — Instantly populate all input fields
🎨 Visual Preview — See template structure before loading
⚡ Quick Access — "Templates" button in action bar
💡 Recommendations — Each template includes actionable tips
🎯 Theme Matching — Auto-suggests templates for your selected theme

Each template includes: Pre-filled context, options, assumptions, risks, evidence fields, AND recommendations.

🖱️ Click "Templates" → 🔍 Search/Filter → 💡 View Tips → 📋 Select Template → ✅ Fields Auto-Filled

🏔️ Decision Landscape (3D Visualization)

Real-time health dashboard showing 6 key decision metrics in a stunning 3D bar visualization:

Metric	Description	Color
📊 Readiness	Overall decision preparedness score	🟢 Green
🔍 Coverage	Blind spot identification coverage	🔵 Blue
📝 Evidence	Quality of supporting evidence	🔷 Cyan
💪 Confidence	Analysis confidence level	🟣 Purple
⚡ Actionable	How actionable the decision is	🩷 Pink
⚠️ Risk Mgmt	Risk management score	🟡 Orange

Visual Features:

🏗️ 3D animated bar towers with gradient lighting
📈 Real-time metric updates as you type
🏥 Overall health percentage with color coding (green/amber/red)
🖱️ Hover interactions for detailed values
✨ Smooth animations and glow effects

📤 Share Reports

Multiple ways to share your decision reports:

Platform	What's Shared
📧 Email	Full PDF report summary with all metrics, opens PDF preview first
💬 WhatsApp	Formatted report with readiness score, grade, completeness
✈️ Telegram	Report summary with key decision metrics

Share includes: Title, Readiness Score, Grade, Completeness, Actionability, and link to full report.

🔤 Text Scramble Effect

Premium animated text reveal for "Grounds" branding:

Feature	Description
🎬 Decode Effect	Characters scramble then resolve to final text
🔄 Auto Loop	Repeats every 8 seconds for eye-catching effect
⚡ Smooth Animation	60ms per character resolution
🎨 Gradient Colors	Emerald-cyan gradient on "grounds" text

📷 Smart Document Scan with Auto-Theme Detection

Upload any image and Grounds automatically extracts decision context:

📷 Upload Image → 🤖 AI Analyzes → 🎯 Theme Detected → 📝 Form Filled → ✅ Ready to Compare

Capability	Description
🔍 Context Extraction	Powered by Llama 4 Scout Vision
🎨 Theme Detection	Auto-identifies industry (Healthcare, Legal, Tech, etc.)
⚡ Instant Fill	All input fields populated automatically
✏️ Editable Results	Review and refine before comparing

Perfect for: 📰 News articles • 📧 Email threads • 📄 Reports • 📱 Screenshots

🎨 11 Industry Themes

Each theme includes curated 2025-2026 case studies based on real-world scenarios:

Theme	The Problem	Example Case Study
🌐 General	Companies facing viral controversies need immediate crisis response strategies	AI consent scandal response - deciding transparency vs. legal defense
💻 Technology	Cloud outages cost billions; AI infrastructure promises resilience but carries lock-in risks	Post-2025 cloud crisis: migrate to AI-autonomous systems or enhance multi-cloud?
🏥 Healthcare	FDA AI approvals create pressure; patients demand AI care but staff fear displacement	Adopting autonomous AI diagnosis after 2025 FDA approval
⚖️ Legal	Supreme Court ruling opened floodgates; firms face 60% price undercuts from AI-powered competitors	AI contract analysis: full adoption vs. premium human-only positioning
💰 Finance	AI-managed portfolios outperformed by 47%; traditional wealth managers losing clients	Launching AI wealth management amid unclear fiduciary regulations
🏛️ Government	Citizens demand smart city benefits but fear surveillance and data misuse	$50M AI infrastructure investment: phased deployment vs. delay
🕊️ Religion & Ethics	2025 "AI Pastor" controversy exposed authenticity crisis in religious leadership	Using AI for sermon research while maintaining spiritual authenticity
📚 Education	78% of students use AI; universities can't distinguish AI-assisted from original work	AI tutors vs. academic integrity: embrace, control, or reject?
🌍 Environment	SEC mandates climate disclosure; greenwashing accusations threaten credibility	AI carbon management: genuine sustainability or compliance theater?
🎬 Media	SAG-AI agreement allows digital twins; AI content costs 70% less but has lower engagement	AI actors: aggressive adoption, hybrid approach, or human-only brand?
🚗 Transportation	Waymo fatality dropped public trust 34%; cities face competing pressures	Autonomous vehicle zones: innovation vs. safety vs. labor concerns

🔍 Multi-Provider AI Comparison

Run your decision through multiple AI providers simultaneously:

Provider	Model	Specialty
⭐ Google Gemini	Gemini 3 Pro/Flash	Featured provider, decision-grade synthesis with multimodal support
🤖 OpenAI	GPT-4o mini	Fast, reliable responses
⚡ Groq	Llama 3.1 8B	Ultra-fast inference (LPU)
🌐 OpenRouter	Multiple	Access to 100+ models

Google Gemini Configuration:

Primary: Gemini 3 Flash (preview) — Optimized for speed
Alternative: Gemini 3 Pro (preview) — Maximum capability
Fallback: Gemini 2.5 Flash — Stable fallback if preview unavailable

Each result displays:

✅ Quality score (0-100)
⏱️ Response time
📊 Token usage
📋 Structured sections detected
🏅 Medal ranking (🥇🥈🥉)

🕵️ Gemini Critic

Powered by Google Gemini 3, the critic performs structural analysis:

Analysis Type	What It Catches
🎯 Hard Edges	Missing numbers, dates, thresholds
💬 Generic Language	Vague phrases that need specifics
👁️ Blind Spots	Missing stakeholders, failure modes
🤔 Assumptions	Things you're taking for granted
✅ Next Actions	Concrete steps with ownership

Critique Score: AI calculates a 0-100 score based on issues found, with interpretation and actionable recommendations.

🎭 Sentiment Analysis (ABSA)

Aspect-Based Sentiment Analysis powered by Llama 4 Scout:

Feature	Description
🎯 Per-Section Analysis	Evaluates Context, Risks, Options, Evidence separately
📈 Confidence Scoring	Percentage-based confidence for each classification
🔑 Key Signal Detection	Identifies words and phrases driving sentiment
🏷️ Sentiment Labels	Positive 😊 • Negative 😟 • Neutral 😐 • Mixed 🤔

Why it matters: Understand how your decision framing is perceived, enabling more balanced stakeholder communication.

🔬 Deep Reasoner: Stress Test

Adversarial AI analysis powered by Gemini 3 that attacks your decision from every angle:

Feature	Description
🛡️ 5-Dimension Robustness	Scores: Robustness, Assumption Strength, Risk Coverage, Evidence Quality, Execution Readiness (0-100 each)
🚨 Critical Flaw Detection	Identifies the single most dangerous weakness with failure scenario
😈 Devil's Advocate	AI-generated adversarial challenges with counter-arguments and suggested responses
🔗 Hidden Dependencies	Finds unacknowledged dependencies that could cause cascading failure
📊 Probability Matrix	Overall failure risk, risk diversity score, single points of failure, correlated risk clusters
🧠 AI Reasoning Path	Step-by-step transparency showing how the AI stress-tested your decision

How it works:

📝 Decision Input → 🔬 Adversarial Analysis → 🛡️ Robustness Scores → 😈 Challenges → 📊 Report

Report Integration:

Results appear in both HTML and PDF reports with plain-language interpretation
Stress test findings are integrated into the Executive Conclusion and Priority Action Checklist
Each score includes an explanation written for non-technical users

Why it matters: Decisions that survive adversarial pressure are more likely to succeed in the real world. This feature ensures you are not surprised by foreseeable failures.

📊 Decision Quality Radar Chart

Six-dimensional visualization of decision quality:

Dimension	What It Measures
📊 Readiness	Overall decision preparedness
⚠️ Risk Coverage	How well risks are identified
📚 Evidence Quality	Strength of supporting data
🧱 Assumption Clarity	How explicit your assumptions are
✅ Actionability	How actionable the decision is
🎯 Confidence	Your certainty level

Features: Interactive tooltips • Color-coded dimensions • Animated rendering

📈 Data Science Visualizations

Six advanced charts designed for analysts and data scientists:

Chart	Purpose	Key Features
📈 Confidence Interval	Quantify uncertainty range	95% CI bands, point estimates, error visualization
📉 Progress Timeline	Track decision quality over time	Trend detection (📈↗️📉), multi-metric tracking
☁️ Word Cloud	Extract key themes from context	Frequency-weighted sizing, theme categorization
🔗 Correlation Matrix	Reveal metric relationships	Heat-mapped coefficients, top correlations highlighted
📦 Box Plot	Analyze provider latency distribution	Quartile analysis, outlier detection, speed ranking
🎲 Monte Carlo Simulation	Probabilistic outcome modeling	1000+ iterations, frequency distribution, confidence bands

Monte Carlo Simulation:

🎲 Probabilistic Modeling — Simulates 1000+ decision outcomes
📊 Frequency Distribution — Histogram showing outcome probability
🎯 Expected Value — Most likely outcome based on simulations
📉 Risk Quantification — Standard deviation and confidence intervals
🔔 Normal Distribution Overlay — Bell curve comparison

All charts feature:

🌙 Dark mode optimized animations
☀️ Light mode optimized animations
🖨️ Perfect PDF export rendering
🎨 Smooth color transitions and glow effects

📝 Executive Conclusion with Action Plan

Comprehensive AI-generated summary with actionable recommendations:

Component	Description
📊 Decision Confidence Score	Weighted percentage (0-100%) combining all metrics
🏷️ Confidence Level	Very High • High • Moderate • Low • Very Low
🎯 Priority Action Checklist	Categorized by urgency (🔴 HIGH • 🟡 MEDIUM • 🟢 LOW)
📋 Key Takeaways	3-5 critical points
⚡ Immediate Next Steps	Concrete actions with ownership
📅 Review Schedule	Suggested follow-up timeline

Confidence Score Formula:

Score = (Readiness × 0.4) + (BlindSpots × 0.3) + (Sentiment × 0.2) + (Baseline × 0.1)

🔬 Related Research

AI-generated research suggestions to inform your decision:

📚 Curated research topics
🔍 Suggested search queries
🏷️ Categorized by type (Research, News, Case Study, Article)
💡 Educational notes for further investigation

📄 Report Generation

Professional reports in multiple formats:

Format	Features
🌐 HTML	Offline-ready, theme-styled, all visualizations, interactive charts
📑 PDF	Print-ready, professional layout, full analytics, perfect rendering
📊 JSON	Structured data export for programmatic access

Report Structure (24 sections):

1.  Header & Metadata          13. Confidence Interval Chart
2.  Executive Brief            14. Progress Timeline
3.  Context & Intent           15. Word Cloud
4.  Options & Trade-offs       16. Correlation Matrix
5.  Assumptions & Risks        17. Box Plot Analysis
6.  Perspective Shift          18. Hackathon Sections
7.  Decision Gauge             19. Provider Compare Details
8.  Radar Chart                20. Related Research
9.  Quality Distribution       21. Grade Legend
10. Provider Spider Chart      22. Data Export (JSON)
11. Sentiment Analysis         23. Executive Conclusion
12. Sentiment Heatmap          24. Disclaimer & Credits

🗂️ Local-First Architecture

Your decisions stay yours:

Feature	Benefit
💾 No Database Required	Runs entirely in browser
📜 Decision Ledger	Timeline of all decisions
🖼️ Gallery View	Browse past reports visually
📤 Export Anytime	Never locked in

💡 Why Use Grounds?

👤 For Individuals

Benefit	Description
🧠 Think Clearer	Structure forces clarity—no more "gut feeling" decisions
⚖️ Compare Perspectives	See how different AI models analyze the same problem
🔍 Find Blind Spots	AI-powered critic reveals what you're missing
📁 Build Your Library	Every decision saved, searchable, reviewable

👥 For Teams

Benefit	Description
📋 Standardize Process	Same framework for every decision
🤝 Align Stakeholders	Shareable reports with clear rationale
📈 Learn Over Time	Review past decisions, improve future ones
⚡ Move Faster	Less meetings, more clarity

🏢 For Organizations

Benefit	Description
✅ Audit Compliance	Full decision trail for governance
🎓 Train Better	Team members learn from decision library
🛡️ Reduce Risk	Catch blind spots before they become problems
💰 Save Money	Better decisions = better outcomes

🎯 Who is This For?

Grounds is built for anyone who makes consequential decisions:

Role	Use Case
👔 Executives	Strategic planning, M&A decisions, resource allocation
💼 Consultants	Client recommendations, project scoping
🏥 Healthcare	Treatment options, policy decisions
⚖️ Legal	Case strategy, settlement decisions
💰 Finance	Investment thesis, risk assessment
🎓 Education	Curriculum changes, policy updates
⛪ Religious Orgs	Ethical decisions, community policies
🚀 Startups	Pivot decisions, hiring, fundraising
📊 Data Scientists	Decision analytics, uncertainty quantification

🛠️ Tech Stack

Layer	Technology	Purpose
Frontend	Next.js 15, React 19, Tailwind CSS	Modern, responsive UI
AI Providers	Google Gemini, OpenAI, Groq, OpenRouter	Multi-provider comparison
Vision AI	Groq Llama 4 Scout	Document scanning, theme detection, sentiment
Scoring	Rust → WASM	Deterministic quality scoring
Report Generation	Python, Playwright	Server-side PDF/HTML report generation
Charts	Custom SVG + CSS Animations	Data visualizations with dark/light mode
State	LocalStorage	Client-side persistence

🔌 API Routes

Endpoint	Description
`/api/compare`	Multi-provider analysis
`/api/gemini-critic`	Structured critique generation
`/api/scan-document`	Image-to-context extraction with theme detection
`/api/sentiment`	Aspect-based sentiment analysis
`/api/conclusion`	Executive summary generation
`/api/related-research`	Research suggestions

🚀 Quick Start

1️⃣ Clone & Install

git clone https://github.com/wiqilee/grounds.git
cd grounds
npm install

2️⃣ Environment Variables

Create .env.local:

# Required: Gemini (Featured Provider + Critic)
GOOGLE_API_KEY=your_google_api_key

# Required: Document Scan, Theme Detection, Sentiment Analysis
GROQ_API_KEY=your_groq_api_key

# Optional: Additional providers
OPENAI_API_KEY=your_openai_api_key
OPENROUTER_API_KEY=your_openrouter_api_key

3️⃣ Run Development Server

npm run dev

Open: http://localhost:3000

4️⃣ Production Build

npm run build
npm start

🧩 How It Works

┌─────────────────────────────────────────────────────────────┐
│                      📷 Smart Scan                          │
│        Upload image → Extract context → Detect theme        │
└─────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────┐
│                     📝 Decision Input                       │
│  Title • Context • Options • Assumptions • Risks • Evidence │
└─────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────┐
│                     🔍 Compare Engine                       │
│   ┌─────────┐  ┌─────────┐  ┌─────────┐  ┌────────────┐   │
│   │Gemini ⭐│  │ OpenAI  │  │  Groq   │  │ OpenRouter │   │
│   └─────────┘  └─────────┘  └─────────┘  └────────────┘   │
└─────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────┐
│                   📊 Analysis Pipeline                      │
│  ┌──────────┐  ┌──────────┐  ┌──────────┐  ┌──────────┐   │
│  │Sentiment │  │  Radar   │  │Conclusion│  │ Research │   │
│  │  (ABSA)  │  │  Chart   │  │Generator │  │  Links   │   │
│  └──────────┘  └──────────┘  └──────────┘  └──────────┘   │
└─────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────┐
│               📈 Data Science Visualizations                │
│  ┌────────┐  ┌────────┐  ┌────────┐  ┌──────┐  ┌───────┐  │
│  │   CI   │  │  Time  │  │  Word  │  │ Corr │  │  Box  │  │
│  │ Chart  │  │ Series │  │ Cloud  │  │Matrix│  │  Plot │  │
│  └────────┘  └────────┘  └────────┘  └──────┘  └───────┘  │
│                      ┌───────────┐                        │
│                      │  Monte    │                        │
│                      │  Carlo    │                        │
│                      └───────────┘                        │
└─────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────┐
│                   🏆 Scoring & Ranking                      │
│        Rust WASM engine → Quality scores → Medal ranks      │
└─────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────┐
│                        📊 Output                            │
│  ┌─────────────┐  ┌─────────────┐  ┌───────────────────┐   │
│  │ HTML Report │  │ PDF Export  │  │  Decision Ledger  │   │
│  └─────────────┘  └─────────────┘  └───────────────────┘   │
└─────────────────────────────────────────────────────────────┘

📚 API Reference

🔍 Compare Endpoint

POST /api/compare
Content-Type: application/json

{
  "prompt": "Decision analysis prompt...",
  "providers": ["google", "openai", "groq"]
}

🕵️ Gemini Critic

POST /api/gemini/critic
Content-Type: application/json

{
  "input": { /* DecisionInput */ },
  "outcome": "Selected outcome text..."
}

📷 Document Scan

POST /api/scan-document
Content-Type: multipart/form-data

file: <image>

Response:

{
  "success": true,
  "detectedTheme": "healthcare",
  "title": "Should we implement AI-assisted diagnosis?",
  "context": "...",
  "options": "...",
  "risks": "..."
}

🎭 Sentiment Analysis

POST /api/sentiment
Content-Type: application/json

{
  "title": "Decision title",
  "context": "...",
  "intent": "...",
  "options": "...",
  "assumptions": "...",
  "risks": "...",
  "evidence": "...",
  "outcome": "..."
}

Response:

{
  "success": true,
  "aspects": [
    {
      "aspect": "Context",
      "sentiment": "neutral",
      "confidence": 78,
      "keySignals": ["factual", "balanced"],
      "summary": "Objective presentation of facts"
    }
  ],
  "overallSentiment": "neutral",
  "overallConfidence": 72,
  "modelUsed": "Llama 4 Scout (Groq)"
}

📝 Executive Conclusion

POST /api/conclusion
Content-Type: application/json

{
  "title": "...",
  "context": "...",
  "readinessScore": 85,
  "grade": "A",
  "confidence": "high",
  "theme": "technology"
}

🔬 Related Research

POST /api/related-research
Content-Type: application/json

{
  "title": "...",
  "context": "...",
  "theme": "healthcare"
}

🔬 Deep Reasoner: Stress Test

POST /api/deep-reasoner
Content-Type: application/json

{
  "title": "Decision title",
  "context": "Background context...",
  "intent": "What success looks like...",
  "options": ["Option A", "Option B"],
  "assumptions": ["Assumption 1", "Assumption 2"],
  "risks": ["Risk 1", "Risk 2"],
  "evidence": ["Evidence 1"],
  "outcome": "Selected outcome...",
  "confidence": "high"
}

Response:

{
  "success": true,
  "result": {
    "scores": {
      "overall_robustness": 62,
      "assumption_strength": 55,
      "risk_coverage": 70,
      "evidence_quality": 48,
      "execution_readiness": 65
    },
    "critical_flaw": {
      "title": "No migration rollback plan",
      "severity": "high",
      "explanation": "If migration fails, there is no documented path to restore previous state.",
      "failure_scenario": "Partial migration corrupts live data with no recovery path."
    },
    "devils_advocate_challenges": [...],
    "hidden_dependencies": [...],
    "probability_matrix": {
      "overall_failure_risk": 38,
      "risk_diversity_score": 52,
      "single_point_of_failures": ["..."],
      "highest_risk_cluster": ["..."]
    },
    "thinking_path": [...],
    "meta": {
      "model_used": "gemini-2.5-flash",
      "processing_time_ms": 4200
    }
  }
}

🧠 Design Philosophy

Principle	Description
🎯 Explicit beats clever	Every decision is visible and auditable
📐 Structure beats persuasion	Frameworks over opinions
⏳ Decisions should age well	Still makes sense 6 months later
🤝 AI assists, human decides	Tools, not oracles
📊 Data-driven insights	Quantify uncertainty, visualize relationships

🚫 What Grounds Is NOT

❌	Description
🚫	Not a legal/financial/medical advisor
🚫	Not a recommendation engine
🚫	Not an "AI decides for you" tool
🚫	Not a chatbot

Grounds is a thinking surface, not an authority.

🌐 Browser Compatibility

Browser	Version
🌐 Chrome	90+ (recommended)
🦊 Firefox	88+
🧭 Safari	14+
🔷 Edge	90+

🤝 Contributing

Contributions welcome! Please read our contributing guidelines.

# Run tests
npm test

# Lint
npm run lint

# Type check
npm run type-check

📄 License

MIT License — Use freely, attribution appreciated.

See LICENSE for full details.

⚠️ Attribution & Ethics Notice

🏆 Competition Submission

This project was built as an original submission for the Google Gemini 3 Hackathon Competition 2026.

While the source code is open under the MIT License, please adhere to the following ethical guidelines:

❌ Do Not	✅ Do
Submit this project or its derivatives as your own in any competition	Fork and build upon it for learning purposes
Claim original authorship in academic contexts	Give proper attribution to the original author
Remove copyright notices or attribution	Keep the LICENSE file and credits intact
Use for misleading or deceptive purposes	Use responsibly and ethically

Attribution is expected when:

Using substantial portions of the codebase
Creating derivative works
Showcasing in portfolios or presentations
Publishing articles or tutorials based on this project

Proper attribution example:

Based on Grounds by Wiqi Lee (https://github.com/wiqilee/grounds)
Original submission for Google Gemini 3 Hackathon Competition 2026

⚠️ Disclaimer

Grounds is an AI-assisted tool designed to support, not replace, human judgment. All analysis, recommendations, and conclusions may contain errors or biases inherent in AI systems. Always verify critical information independently and consult domain experts for high-stakes decisions.

👨‍💻 Built by Wiqi Lee

Platform	Handle
🐦 Twitter	@wiqi_lee
💻 GitHub	wiqilee
💬 Discord	wiqi_lee

⭐ Star this repo if you find it useful!

🏆 Built for Google Gemini 3 Hackathon 2026

Name		Name	Last commit message	Last commit date
Latest commit History 14 Commits
app		app
components		components
engine		engine
lib		lib
report		report
rust/score_engine		rust/score_engine
sample		sample
scripts		scripts
.gitignore		.gitignore
LICENSE		LICENSE
README.md		README.md
next-env.d.ts		next-env.d.ts
next.config.mjs		next.config.mjs
package-lock.json		package-lock.json
package.json		package.json
postcss.config.mjs		postcss.config.mjs
tailwind.config.ts		tailwind.config.ts
tsconfig.json		tsconfig.json

License

wiqilee/grounds

Folders and files

Latest commit

History

Repository files navigation

Grounds ✨

🚀 Live Demo

▶️ Video Demo

🎯 What is Grounds?

⚡ The Problem

💡 The Solution

✨ Features

🎤 Voice Input (Multimodal)

📖 About & Guided Tour

📤 Google Workspace Export

🔍 Gemini Web Search Integration (Grounding)

📋 Decision Templates & Template Gallery

🏔️ Decision Landscape (3D Visualization)

📤 Share Reports

🔤 Text Scramble Effect

📷 Smart Document Scan with Auto-Theme Detection

🎨 11 Industry Themes

🔍 Multi-Provider AI Comparison

🕵️ Gemini Critic

🎭 Sentiment Analysis (ABSA)

🔬 Deep Reasoner: Stress Test

📊 Decision Quality Radar Chart

📈 Data Science Visualizations

📝 Executive Conclusion with Action Plan

🔬 Related Research

📄 Report Generation

🗂️ Local-First Architecture

💡 Why Use Grounds?

👤 For Individuals

👥 For Teams

🏢 For Organizations

🎯 Who is This For?

🛠️ Tech Stack

🔌 API Routes

🚀 Quick Start

1️⃣ Clone & Install

2️⃣ Environment Variables

3️⃣ Run Development Server

4️⃣ Production Build

🧩 How It Works

📚 API Reference

🔍 Compare Endpoint

🕵️ Gemini Critic

📷 Document Scan

🎭 Sentiment Analysis

📝 Executive Conclusion

🔬 Related Research

🔬 Deep Reasoner: Stress Test

🧠 Design Philosophy

🚫 What Grounds Is NOT

🌐 Browser Compatibility

🤝 Contributing

📄 License

⚠️ Attribution & Ethics Notice

⚠️ Disclaimer

👨‍💻 Built by Wiqi Lee

About

Topics

Resources

License

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Languages

Packages