KumoKodo.ai Case Study
Building SiteBeacon
How we created an AI-powered lead generation platform that helps web agencies find businesses in need of their services—using Google Maps, Gemini AI, and modern SaaS architecture.
3
Months Development
15K+
Lines of Code
40+
API Endpoints
100%
TypeScript
Client
SiteBeacon
Industry
B2B SaaS / Lead Generation
Services
Full-Stack Development
The Challenge
Web developers and digital agencies face a critical problem: finding qualified leads is time-consuming, expensive, and often ineffective. The traditional approach of cold calling, buying outdated lead lists, or relying solely on referrals doesn't scale.
Pain Points Identified
Goals Defined
The Solution
KumoKodo.ai designed and built SiteBeacon from the ground up—a full-stack SaaS platform that combines Google Maps intelligence with Gemini AI to automate lead discovery, qualification, and outreach.
Multi-Mode Search Engine
Built a sophisticated search system with three distinct modes: Website (finds businesses without websites or using third-party platforms), Reputation (discovers low-rated businesses), and GMB (identifies incomplete Google Business Profiles). Each mode uses specialized filtering algorithms.
AI-Powered Analysis
Integrated Google Gemini 2.0 Flash for instant lead analysis, personalized outreach email generation, and competitor intelligence. The AI understands business context and generates highly relevant, conversion-optimized content using structured JSON schemas.
Intelligent Lead Scoring
Developed a proprietary scoring algorithm (1-100 scale with A-F grades) that evaluates businesses based on website status, rating, review count, GMB completeness, and detected issues. Higher scores indicate better opportunities for conversion.
Team Collaboration
Built enterprise-ready team features with role-based access control, shared credit pools, centralized lead management, email invitations, and owner/member permissions. Perfect for agencies with multiple sales reps.
Technical Architecture
Frontend Stack
Backend Stack
Google Maps Platform
Real-time business data
Places API (New)
Search businesses by location with comprehensive filtering and up to 60 results via pagination.
Place Details API
Retrieve ratings, reviews, websites, phone numbers, hours, photos, and GMB status.
Geocoding API
Convert ZIP codes to lat/lng coordinates for precise radius-based searches.
Google Gemini AI
Advanced language model
Lead Analysis
AI generates comprehensive business summaries highlighting problems and opportunities.
Email Variants & Sequences
Generates multi-variant outreach emails and timed follow-up sequences for each prospect.
Competitor Intelligence
Analyzes local competitors to identify market opportunities and talking points.
Website Audit
PageSpeed-powered technical audit with AI-generated improvement recommendations.
Stripe Billing
Subscriptions & Credit Packs
NextAuth.js
OAuth & Credentials
MongoDB Atlas
Scalable Database
Vercel Edge
Global Deployment
Key Features Delivered
Multi-Mode Search
Three specialized search modes to find exactly the leads you need based on your services.
AI-Powered Outreach
Gemini 2.0 Flash generates personalized content based on each business's specific situation.
Lead Scoring System
Proprietary algorithm evaluates and prioritizes leads so you focus on the best opportunities.
Team Collaboration
Enterprise-ready features for agencies and sales teams of any size.
Flexible Billing
Credit-based system with subscriptions and one-time purchases to fit any budget.
Lead Management CRM
Lightweight CRM features built right into the platform for pipeline tracking.
The Results
17+
Third-Party Platforms Detected
60
Leads Per Search (Max)
<2s
AI Response Time
99.9%
Uptime (Vercel SLA)
For Users
For Business
Lessons Learned
Start with the AI, not the UI
We built the Gemini AI integration and lead scoring algorithm first, then designed the interface around them. This ensured AI was a core feature, not an afterthought bolted on at the end.
Credit systems need atomic transactions
Every credit deduction uses Prisma transactions to prevent race conditions. This was critical for the team credit pool feature where multiple users might search simultaneously.
Third-party detection is surprisingly complex
Businesses use dozens of platforms beyond just "no website." We built detection for 17+ platforms including Facebook, Yelp, DoorDash, Toast, and more—each requiring URL pattern matching.
Type everything from day one
100% TypeScript with strict mode caught countless bugs before they reached production. Combined with Prisma's type-safe queries, we have end-to-end type safety from database to UI.
About KumoKodo.ai
We specialize in building AI-powered SaaS applications using modern cloud-native technologies. Our team combines expertise in machine learning, full-stack development, and user experience design to create products that solve real business problems.
Want a Project Like This?
KumoKodo.ai specializes in building production-ready AI applications and SaaS platforms. Let's discuss your next project.