This content originally appeared on DEV Community and was authored by depa panjie purnama
This is a submission for the AI Agents Challenge powered by n8n and Bright Data
What I Built
I created an AI-powered Company Intelligence Bot that transforms simple company names into comprehensive business intelligence reports delivered instantly via Telegram. This workflow combines the power of Bright Data’s web scraping capabilities with AI analysis to provide investors, entrepreneurs, and market researchers with instant, actionable insights about any company.
The bot solves the problem of time-consuming manual research by automatically analyzing companies across 20+ metrics in five key categories: Business Metrics (market value, growth rate, profitability), Investment Profile (funding score, ROI potential, risk assessment), Innovation & Tech (innovation index, R&D investment), Team & Culture (leadership quality, employee satisfaction), and Market & Competition (market size, competitive advantage).
What makes this solution unique is its ability to handle both startups and Fortune 500 companies with equal precision, automatically adjusting its analysis based on company maturity, public/private status, and available data points. Users simply send a company name to the Telegram bot and receive a visually rich analysis with progress bars, star ratings, and AI-generated executive summaries within seconds.
Demo
The workflow is deployed and accessible via Telegram bot. Simply send any company name to the bot, and within seconds, receive a comprehensive analysis with visual progress bars, star ratings, and AI-generated executive summaries.
Screenshot showing real-time analysis of Amazon with comprehensive metrics and visual indicators
The demo showcases instant response times, visual analytics with progress bars and color-coded status indicators (
), overall company ratings on a 5-star scale, AI-generated executive summaries, and conversation memory that maintains context across multiple queries for deeper analysis.
n8n Workflow
The complete workflow JSON is available here: GitHub Gist – Company Intelligence Bot Workflow
The workflow consists of 8 interconnected nodes: Telegram Trigger (receives company name queries), URL Builder (constructs Crunchbase URLs dynamically), Bright Data Scraper (fetches real-time company data), Rating Calculator (processes 20+ metrics across 5 categories), AI Agent (generates contextual insights using Google Gemini), Memory Buffer (maintains conversation context), Output Formatter (creates visually rich Telegram messages), and Response Sender (delivers formatted analysis to users).
Technical Implementation
The AI agent is configured with Google Gemini Chat Model for natural language processing, providing concise executive summaries based on the analyzed company data. The system uses Buffer Window Memory with 10-message context retention, enabling follow-up questions and maintaining conversation flow. Each Telegram chat receives a unique session ID for personalized interactions.
The workflow implements a sophisticated multi-stage pipeline: First, the Data Acquisition Layer utilizes Bright Data’s verified node to scrape Crunchbase data, handling various company formats and implementing error handling for pending snapshots. Next, the Analysis Engine uses custom JavaScript algorithms to calculate ratings across 5 dimensions with adaptive scoring that adjusts based on company type. The AI Enhancement Layer then processes this data through Google Gemini with specific instructions to generate professional summaries. Finally, the Visualization Engine creates dynamic progress bars with emoji indicators, star ratings with half-star precision, and formatted currency displays.
Tools used include n8n’s Telegram nodes for bot interaction, Code nodes for data processing and rating calculations, Bright Data’s verified node for web scraping, n8n’s AI Agent with Google Gemini integration, and Memory nodes for conversation persistence.
Bright Data Verified Node
The Bright Data integration is the cornerstone of this workflow, configured to use the Crunchbase companies information dataset. The node dynamically constructs URLs based on company names received from Telegram, automatically formatting them for Crunchbase compatibility. It extracts comprehensive data points including company fundamentals (name, founded date, headquarters, website), financial metrics (funding rounds, total raised, valuation, market cap), organizational data (employee count, leadership, investor information), and market positioning (categories, rank, IPO status).
Bright Data was essential for this project because it provides reliable, consistent data extraction from Crunchbase’s complex structure, sub-second response times for cached data, access to both public and private company information, and the ability to handle high-volume requests without rate limiting. The verified node eliminates the complexity of web scraping, allowing me to focus on analysis algorithms and user experience rather than data extraction challenges. The error handling gracefully manages pending snapshots and data unavailability, ensuring users always receive meaningful responses even when data is being processed.
Journey
Creating a tool that could analyze any company – from a 2-person startup to Amazon – required solving several complex challenges. The first major challenge was data normalization: different companies have vastly different data availability (public companies have market caps while startups have funding rounds). I solved this by creating an adaptive scoring system that intelligently adjusts based on available data points, using conditional logic to apply different calculation methods for public vs. private companies.
The second challenge was achieving real-time performance in a messaging app where users expect instant responses. I optimized the workflow by implementing smart caching strategies, parallel processing where possible, and graceful degradation for missing data. The Bright Data node’s speed was crucial here, providing near-instant data retrieval for most queries.
The third challenge was visual communication – presenting complex financial data in a mobile-friendly Telegram format. I developed a custom progress bar system using Unicode characters (▓░), created an emoji-based status indicator system for quick visual assessment, and implemented responsive formatting that works across all devices.
The most significant technical breakthrough was the adaptive rating algorithm that automatically calibrates based on company type. For example, the market value calculation uses different logic for public companies (based on market cap) versus startups (based on funding and valuation). The system also implements intelligent error handling for edge cases like pending data snapshots, companies not found on Crunchbase, and incomplete data profiles.
Key lessons learned include: Data quality matters – Bright Data’s verified node provided consistent, high-quality data that would have been impossible with custom scrapers. User experience is paramount – the visual progress bars and emoji indicators transformed dry financial data into engaging content. The combination of AI and real-time web data creates insights neither could provide alone. Building an adaptive system that handles diverse company types made the tool universally useful.
This project demonstrates how n8n’s workflow automation, Bright Data’s reliable web scraping, and AI-powered analysis can create an unstoppable workflow that democratizes access to business intelligence, enabling investors to quickly screen potential investments, entrepreneurs to research competitors, job seekers to evaluate employers, and researchers to gather company data efficiently.
This content originally appeared on DEV Community and was authored by depa panjie purnama