Web app hosted on Replit. The goal is to help users find relevant podcasts and YouTube channels for guest appearances or collabs, then manage outreach in one place. MUCH OF THIS IS ALREADY BUILT - It just needs to be tested and the database system set up correctly. Users enter a topic or niche, the app should run “deep AI searches” to find shows, pull contact info, store everything in a database, and later use that for outreach and follow-ups. What the app is supposed to do User side Sign up / log in securely Choose niche, language, platform (podcasts / YouTube, etc.) Run an AI-powered search to: Discover relevant shows and channels Extract contact info (emails, websites, socials, host name) Save these results into a shared database so the app can reuse them (no repeated scraping if they already exist). Eventually: let users send outreach emails and follow-ups. Admin / Backend side Manage users and security Control OpenAI API key usage (token limits, etc.) Manage pricing and subscription plans View and manage the database of shows, contacts, and searches Payments / Checkout Streamlined checkout flow for subscriptions (Stripe + PayPal) Monthly and yearly plans (with a yearly discount) After successful payment, automatically upgrade the user’s account What is not working right now 1. AI search / discovery flow The “deep AI search” step is unreliable: It often returns no or very few results when it should find many. Relevance is poor – shows that don’t match the niche or topic. Sometimes it fails silently or errors out on Replit. The chain from: user query → AI → external search (web/APIs) → cleaned show list is not stable or not accurate yet. 2. Contact extraction + database saving Even when shows are found, contact info often: isn’t extracted correctly from descriptions/websites is missing even when emails are clearly present on the channel/page Data is not being saved reliably into the database: Some results never get stored. Duplicates aren’t being handled (same show added multiple times). The structure for shows / hosts / contacts is not clearly enforced. 3. Database integration I need a solid schema and working code path that: Creates or updates a shows table (with platform, URL, title, niche, etc.) Creates or updates a contacts table (email, host name, notes, etc.) Links shows ↔ contacts properly The app should always: Run the search Normalize the results Save them to the DB Then display them to the user Right now, that pipeline is broken or incomplete. 4. Signup, security, and checkout flow Signup/login works but needs to be cleaner and more secure: Proper password handling, sessions, basic protections Clear user roles (admin vs normal user) Checkout/payment is clunky: I want a smooth subscription checkout flow. Pricing needs to be editable from an admin area. After payment, user access level should update automatically. What I need from a developer Review the existing codebase on Replit and map the current flow (auth → AI search → scraping/APIs → DB → UI → payment). Fix or redesign the AI search pipeline: Robust way to search podcasts/YouTube (via APIs + AI). Reliable parsing and cleaning of results. Implement reliable contact extraction: Pull emails and relevant links from channel pages/descriptions/websites. Implement/repair the database layer: Proper schema for shows, contacts, users, searches. Ensure all successful searches are saved. Handle deduplication. Clean up signup/login and security. Streamline the checkout/subscription flow (Stripe & PayPal). Do a live test with me (end-to-end: new user → search → DB → checkout). If you’re comfortable with: AI integrations (OpenAI or similar), calling external APIs (YouTube, podcast directories, etc.), building/cleaning data pipelines, and setting up subscriptions/auth on a Replit-hosted app,