📦 Bidi – Smart Bid Deals Notifier

Bidi is a Telegram-based system that helps Israeli shoppers save big on tech products by revealing hidden bid-based deals from online retailers like ALM, KSP, and more.

Some tech retailers quietly allow you to “Make an Offer” on products — but very few people know about it.
Bidi monitors these deals, estimates the most likely accepted bid, and alerts users in real-time — either in a public Telegram channel or via a personalized Telegram bot experience.


🧠 Why This Matters

This is a new concept for Israeli e-commerce. Most users aren't aware they can bid on products — and even fewer know how much to offer.
Bidi makes it simple, smart, and accessible — for free.


✅ Core Features

🔔 Telegram Channel

  • Public channel that posts hot bid-based deals with product details, store link, and estimated accepted price.

🤖 Telegram Bot (1-on-1)

  • Send /deal to get a random active deal.
  • Set your preferred categories and minimum discount via bot settings.
  • Get notified directly in the bot when new relevant deals go live.
  • /unsubscribe to stop all notifications at any time.

📦 Deal Scraper

  • Scrapes Israeli tech stores (starting with ALM, KSP in progress).
  • Extracts products with active bid options.
  • Estimates a “smart bid” price based on product data and pricing history.

📊 Admin Tools (WIP)

  • Dashboard for internal monitoring and possibly premium users.
  • Filter, sort, and track product performance, bid success rates, and more.

🛠️ Built With

  • SvelteKit – my favorite fullstack framework
  • Supabase – database, auth, storage, Edge Functions, cron jobs
  • Grammy – Telegram bot framework
  • Fly.io + Typesscript + Hono – scraper microservice
  • Cloudflare Workers – secure hosting

⚙️ Architecture Notes

  • Supabase Edge Functions + cron jobs handle periodic tasks like updating product data and syncing stats.
  • For heavier scraping jobs, a Hono-based microservice is deployed on Fly.io, since platforms like Cloudflare Workers or Supabase Edge Functions timeout after ~30 seconds.
  • The system is modular — new retailers can be added with custom scrapers while keeping everything categorized under a unified structure.

🧩 Challenges

  • Each retailer uses a different structure (REST, GraphQL, or no API at all).
  • Some require headless browser automation to simulate bidding (e.g., Playwright).
  • Creating a normalized category tree from multiple sources was non-trivial — needed to unify disjointed product hierarchies.
  • Telegram bots and channels required balancing between broadcast content and personal, interactive flows.

🧪 What's Working Now

  • ALM integration is fully live and pushing real bid-based deals.
  • KSP integration is in progress.
  • Telegram channel + bot deployed and active.
  • Landing page with multilingual support.
  • Serverless backend infrastructure fully automated with cron-based scraping.

💡 What’s Next

  • Add more stores (e.g., BUG, Ivory).
  • Full dashboard for power users with:
    • Price history
    • Filtering by brand/category
    • Product tracking & alerts
    • Exporting data
  • Add monetization:
    • Affiliate integrations (KSP already supports it)
    • Premium dashboard tier

🌍 Why It’s Different

Bidi blends smart automation, scraping, and serverless infra into a single tool that helps real users save money on tech — in a market where this type of deal discovery doesn’t exist yet.

If you've ever overpaid for tech in Israel — you probably could’ve bidi’d it.

Built With

Share this project:

Updates