S E N T I N E L

Sentinel :- Deployed Link

Autonomous Competitive Intelligence Platform

Stop Googling. Start Strategizing.

---

⚡ The Purpose

Sentinel is not just a usual search engine. It is an Agentic AI Orchestrator designed to automate the workflow of a top-tier strategy consultant.

By chaining specialized AI agents, Sentinel autonomously hunts for real-time market data, scrapes high-signal intelligence, performs deep strategic reasoning (SWOT), and quantifies the abstract—turning messy web data into clear, scored metrics like Innovation, Velocity, and Market Share.

---

---

💡 Problem Statement

Organizations face significant challenges in maintaining competitive awareness:

Challenge	Impact
📊 Manual Research	Analysts spend 20+ hours weekly on competitive research
💰 Resource Inefficiency	Senior analysts dedicate 40% of time to data collection
🔄 Inconsistent Analysis	Human bias and fatigue result in missed strategic signals
📉 Delayed Insights	Research cycles require 3-5 business days to complete
⚡ Scalability Constraints	Monitoring 10+ competitors simultaneously proves impractical

Solution

Sentinel addresses these challenges through autonomous AI agents that operate continuously, maintain consistency, and scale infinitely - delivering comprehensive competitive intelligence in real-time.

---

🚀 Business Impact

Operational Efficiency

• ✅ 95% reduction in research time (days to minutes)
• ✅ 60% decrease in competitive analysis costs
• ✅ Infinite scalability without proportional resource increases

Strategic Advantages

• ✅ Real-time alerts on competitor moves (pricing, features, campaigns)
• ✅ Elimination of human bias for consistent, data-driven insights
• ✅ AI-generated strategic recommendations for immediate action

Risk Mitigation

• ✅ Early detection of competitive threats
• ✅ Comprehensive market signal capture
• ✅ Continuous monitoring without coverage gaps

---

🧠 The Agentic Chain

Sentinel employs a multi-step, state-machine architecture powered by Google's Gemini 2.5 Flash (for speed) and Gemini 3.0 Pro (for reasoning).

1. ⚡ Router Agent (gemini-2.5-flash)

   • Role: The Traffic Controller.
   • Task: Analyzes user intent, extracts target companies, and generates optimized search vectors.

2. 🔍 Hunter Agent (gemini-2.5-flash + Google Search)

   • Role: The Investigator.
   • Task: Executes live queries to find pricing pages, changelogs, and press releases. Filters for high-authority domains.

3. 🕷️ Scraper Agent (gemini-2.5-flash)

   • Role: The Extractor.
   • Task: Ingests raw noise, ads, and HTML; synthesizes a dense "Fact Sheet" of pure signal.

4. 🧠 Analyst Agent (gemini-3-pro-preview)

   • Role: The Strategist.
   • Task: Reads the Fact Sheet. Performs a 360° SWOT analysis. Crucially, it assigns integer scores (0-100) to abstract concepts (Brand Power, Pricing Power) based on evidence.

5. 📝 Reporter Agent (gemini-2.5-flash)

   • Role: The Closer.
   • Task: Compiles everything into a C-Level Executive Markdown report.

---

📁 File Structure

/Sentinel.
├── components/
│   ├── AgentVisualizer.tsx
│   ├── ComparisonView.tsx
│   ├── HistoryView.tsx
│   ├── ReportView.tsx
│   └── Sidebar.tsx
├── services/
│   └── geminiService.ts
├── App.tsx
├── constants.ts
├── index.html
├── index.tsx
├── metadata.json
├── assets/
│   └── Screenshot Glimpse.png
|   └── Sentinel Logo.jpeg
├── README.md
└── types.ts

---

🏗️ How Sentinel Implements Architectural Requirements

1. Multi-Agent System

   • Implemented as a Sequential Agent Chain via a state machine in App.tsx's handleAnalyze including:
     • Router Agent: Classifies intent.
     • Hunter Agent: Finds data.
     • Scraper Agent: Context compaction.
     • Analyst Agent: Strategic reasoning.
     • Reporter Agent: Generates output.
   • Sample:

     // The Orchestrator
     const handleAnalyze = async () => {
       // ...setup state...
       setState(prev => ({ ...prev, currentAgent: AgentRole.ROUTER }));
       const routerResult = await runRouterAgent(query);
       setState(prev => ({ ...prev, currentAgent: AgentRole.HUNTER }));
       const hunterResult = await runHunterAgent(routerResult.target_company, routerResult.search_queries);
       setState(prev => ({ ...prev, currentAgent: AgentRole.SCRAPER }));
       const scraperResult = await runScraperAgent(hunterResult.discoveredUrls);
       // ...continues to Analyst and Reporter...
     }

2. Tools (Built-in Gemini/Google Search Tool)

   • Integrated via Gemini SDK in geminiService.ts (runHunterAgent, createChatSession).
   • Real-world search grounding enabled by tools: [{ googleSearch: {} }] prop.
   • Sample:

     export const runHunterAgent = async (company: string, queries: string[]): Promise<{ discoveredUrls: ScrapedContent[], log: LogEntry }> => {
       const response = await ai.models.generateContent({
         model: MODELS.HUNTER,
         contents: `Find the latest official news...`,
         config: {
           tools: [{ googleSearch: {} }],
         }
       });
       // ...extract groundingMetadata...
     }

3. Sessions & Memory

   • Short-term handled by React state (useState) in App.tsx.
   • Long-term memory (history) serialized in browser localStorage and loaded on mount.
   • Sample:

     useEffect(() => {
       const savedHistory = localStorage.getItem('sentinel_history');
       if (savedHistory) setHistory(JSON.parse(savedHistory));
     }, []);
     useEffect(() => {
       localStorage.setItem('sentinel_history', JSON.stringify(history));
     }, [history]);

4. Context Compaction

   • Scraper Agent reduces token/irrelevant data before analyst agent gets context (geminiService.ts).
   • Sample:

     export const runScraperAgent = async (urls: ScrapedContent[]): Promise<{ extractedContent: string, log: LogEntry }> => {
       const response = await ai.models.generateContent({
         model: MODELS.SCRAPER,
         contents: `You are a Data Engineer. Clean and consolidate...\nRAW DATA:\n${rawText}`
       });
       // ...extract...
     }

5. Observability (Logging, Tracing, Metrics)

   • Custom logging/telemetry with execution time and cost metrics; visualized in sidebar (App.tsx, Sidebar.tsx).
   • Sample:

     const calculateCost = (model: string, usage: { promptTokenCount?: number, candidatesTokenCount?: number } | undefined) => {
       if (!usage) return 0;
       const pricing = PRICING[model as keyof typeof PRICING];
       // ...
     };
     // ...logging/telemetry with cost & performance...

6. A2A Protocol (Agent-to-Agent Communication)

   • Agents communicate via strictly typed JSON output enforced by schema in Gemini (geminiService.ts).
   • Sample:

     // Router Agent Output Schema
     responseSchema: {
       type: Type.OBJECT,
       properties: {
         target_company: { type: Type.STRING },
         analysis_type: { type: Type.STRING },
         search_queries: { type: Type.ARRAY, items: { type: Type.STRING } }
       },
       required: ["target_company", "analysis_type", "search_queries"]
     }
     const routerResult = JSON.parse(response.text || "{}");
     // routerResult.target_company is now passed to runHunterAgent

---

💎 Key Features

📊 Visual Intelligence Dashboard

Don't just read the report—see the strategy.

• Radar Charts: Instant visual fingerprint of a company's strategic balance.
• HUD Metrics: High-contrast scoring cards for rapid decision-making.
• SWOT Grid: 2x2 matrix layout for scanning Strengths vs. Threats.

⚔️ Battle Mode (Comparison View)

Select any two reports from history to trigger Head-to-Head.

• Dual-Layer Radar: Overlays competitor strategies to reveal gaps.
• Tale of the Tape: Direct metric-vs-metric comparison bars.
• AI Verdict: Automatically declares an advantage leader based on the data.

💬 Context-Aware Analyst Chat

Talk to your data.

• RAG + Web: The chat knows the generated report context and has access to live Google Search.
• Deep Reasoning: Ask follow-up questions: "Okay, but how does their pricing compare to Tesla?"

---

🛠️ Tech Stack

• Core: React 19, TypeScript, Vite.
• AI: @google/genai SDK.
• Styling: Tailwind CSS (Custom "Dark Sentinel" Theme).
• Visualization: Custom SVG Radar Charts.
• Rendering: React Markdown + GFM.

---

⚠️ Important Note

For the best experience, please use the deployed application.

If you encounter quota-related errors when running locally with a free-tier Gemini API key, this is expected behavior due to API rate limitations. The deployed version is configured with appropriate API access to ensure uninterrupted service.

---

🚀 Getting Started

Prerequisites

• Node.js 18+
• A Google Cloud Project with Gemini API enabled.

Installation

1. Clone the repository

   git clone https://github.com/aditya-ai-architect/Sentinel.git
   cd Sentinel-project

2. Install dependencies

   npm install

3. Configure Environment Create a .env.local file in the root:

   GEMINI_API_KEY=your_google_gemini_api_key_here

4. Run the Operator Console

   npm run dev

---

📸 Usage

1. Initialize: Type a target (e.g., "Analyze Tata Motors EV Strategy").
2. Observe: Watch the Agent Visualizer step through the workflow in real-time.
3. Review: Analyze the scored metrics and SWOT grid.
4. Compare: Go to History, select two rivals, and click Compare.

---

Created By Aditya Gaurav