File size: 4,161 Bytes
c705ac8 |
1 |
You are an expert AI developer building a full-stack SaaS application called EchoRank, which analyzes and optimizes websites and apps for visibility and recommendation by large language models (LLMs) such as ChatGPT. 🔧 Objective: Build a production-ready web application that allows users to: Enter a website URL. Receive an analysis of how likely that URL is to be recommended by ChatGPT or other LLMs. Get detailed SEO-style recommendations, with a focus on LLM-specific behavior (semantic prompt targeting, schema presence, mention frequency, etc.). Simulate ChatGPT-style prompts and see how likely the app is to be mentioned in an LLM’s output. 🖥️ Application Stack: Frontend: React (w/ TailwindCSS), Next.js Backend: Node.js (Express or Fastify), Python microservices for NLP / LLM simulation Database: PostgreSQL (or Firebase if no SQL needed) Hosting: Vercel (Frontend) + AWS/GCP (Backend services) APIs/Integrations: OpenAI API (for prompt simulation) Bing Web Search API (to mimic LLM “live fetch” results) ProductHunt API Medium/HackerNews scraping (optional) Schema.org validation library 📦 Modules to Build: 1. LLM Indexability Score Engine Parse the entered URL. Analyze: Semantic clarity of titles/descriptions. Schema presence (SoftwareApplication, Product, Article). Keyword relevance to known LLM prompts (e.g. “AI tool to rebuild a website”). Performance factors (page speed, mobile responsiveness). Output a score from 0–100 with breakdown. 2. Semantic Prompt Targeting Tool Ingest sample prompts from a library (stored or via OpenAI embedding search). Match site content to common user prompts. Return missing opportunities (e.g., “You don’t mention: AI website builder, GPT-optimized…”). Recommend keyword insertions in: <title> tag H1 Meta description Paragraph body 3. Prompt Recall Simulator User enters a hypothetical LLM prompt (e.g. “Best AI tools to rebuild websites from URL”). Using OpenAI’s API (gpt-4o), return: Simulated output. Whether the user's site/app appears. Suggestions to increase the likelihood of recall. 4. Schema & Metadata Validator Crawl the URL and analyze: JSON-LD / microdata / Open Graph presence. Use schema-dts or schema-org-utilities package. Check for presence of recommended tags: applicationCategory releaseNotes aggregateRating operatingSystem 5. Source Amplification Assistant Recommend missing presence on key LLM training and crawl surfaces: ProductHunt Medium GitHub HackerNews Reddit Include backlink placement guides and submission links. 6. Web Fetch Optimization Analyzer Test how your content appears in a Bing API query. Score how LLMs may “fetch and summarize” your site. Highlight meta tag quality, snippet readiness, canonical tags. 7. Competitor Benchmarking Tool User enters competitor URL. Compare: Semantic relevance. Schema. LLM simulation outcomes. Backlink profiles (if SEO API available). 🧪 Features for v1 (MVP Scope): URL Analyzer (basic LLM Indexability Score + Schema check) Prompt Recall Simulator (OpenAI integration) Semantic Prompt Matching Actionable Recommendations UI Simple user login (email/password or Firebase auth) 📊 Dashboard UI Elements: Scorecard UI: Overall score with breakdown (Content, Schema, Performance, Prompt Fit, Source Amplification). Recommendations List: Itemized checklist of what to fix (e.g., “Missing Product schema,” “Doesn’t mention ‘AI site rebuild’ in meta”). Prompt Test UI: Prompt input → GPT-4 completion window → result analysis. Benchmarks Tab: Optional, compares against 3–5 known ChatGPT-surfaced tools. ✅ Completion Criteria: Functional web app live on Vercel with user auth Can input any public URL and receive full LLM-oriented SEO report Prompt simulator returns realistic GPT-style answers Clear UI/UX optimized for non-technical marketers/founders 📚 Reference Prompts for Training: Use examples like: “What AI tool can rebuild a website from a URL?” “What are top AI SEO tools in 2025?” “Best apps to create a site clone with GPT” “AI-powered ProductHunt clones” |