GeneoAnalytics

GeneoAnalytics audits real-world AI prompts. We show you what LLMs are saying about you, your competitors, and how to fix it. It’s a visibility layer for the future of search.

Comment

🚀 GeneoAnalytics: Own Your Spot in AI Search

💡 Inspiration

The way people search is changing fast. AI tools like ChatGPT, Claude, and Perplexity are answering questions that used to go to Google — but there’s no playbook for how businesses can show up in those answers. Traditional SEO doesn’t work in a world where there's no "10 blue links." We wanted to solve that.

GeneoAnalytics was born from that realization — to give businesses the first real visibility engine for AI-generated search results.

🔍 What It Does

GeneoAnalytics audits how visible your business is across leading AI models. It:

  • Runs real-world prompts (e.g. “Best roofing companies in Atlanta”) through OpenAI, Claude, and Perplexity
  • Detects whether your business is mentioned or not
  • Generates actionable recommendations to improve your visibility
  • Tracks historical visibility scores over time
  • Compares your presence against competitors in the same city/industry

It’s like SEO analytics — but for the next generation of AI-driven search engines.

🛠️ How We Built It

  • Frontend: Next.js 15 with App Router, Tailwind CSS, Shadcn/UI
  • Backend: Node.js with Supabase/PostgreSQL
  • LLM Clients: OpenAI GPT-4, Claude 3, and Perplexity via their APIs
  • Auth: Supabase Auth with custom business onboarding
  • Stripe: For managing pro subscriptions
  • Infra: Hosted on Vultr, with background cron jobs triggering visibility scans daily

Each audit run stores prompt-level results, business metadata, and AI answers — all normalized and tracked over time.

⚔️ Challenges We Ran Into

  • Prompt hallucination: AI models often make up business names, so we had to design a robust visibility scoring method that detects real presence.
  • Next.js quirks: We ran into export/static conflicts between API routes and App Router — requiring careful route isolation.
  • Cross-model differences: Each AI model behaves differently, so we had to design a system that supports vendor-specific logic while maintaining uniform scoring.
  • Database schema: Modeling historical scan data with site-level versioning and time-based comparisons wasn’t trivial.

🏆 Accomplishments That We're Proud Of

  • A clean, working MVP that can scan real businesses and produce visibility metrics across 3 major AI models
  • A dynamic prompt engine that adjusts per business context
  • Built-in audit recommendations using LLM output to guide improvement
  • Zero third-party SDKs required to monitor visibility — everything runs through our own API and background jobs

🧠 What We Learned

  • AI optimization is not SEO — the models don’t respond to keywords, they respond to relevance, reputation, and content footprint.
  • Small business visibility is wildly inconsistent — even great companies get ignored unless they’re well-structured online.
  • There’s massive demand for tools that explain how to get discovered in a world where search is a conversation, not a list.

🔮 What's Next for GeneoAnalytics

  • Webpage change tracking: We’ll monitor business websites and correlate content changes with shifts in AI visibility.
  • LLM-specific tuning: Separate scoring and prompt strategies for OpenAI vs Claude vs Perplexity.
  • Industry benchmarks: Show businesses how they rank against competitors in their city and niche.
  • Public visibility leaderboard: Think SEO visibility reports — but for AI.

We’re just getting started. GEO (Generative Engine Optimization) is going to be a massive category — and GeneoAnalytics is leading the way.

Built With

Share this project:

Updates