← sources

Wrote a program to detect e-commerce price drops & loot deals, and auto-categorize them into clean feeds

source: reddit (reddit:developersIndia)published 2026-10-13 16:46ingested 2026-10-14 02:37

retained text

Every festive sale season (big billion days / great indian festival), Telegram deal groups become unusable. They blast 500+ messages a day, but only about 5% are actual 70-80% price glitches. The rest is just random clothes, phone cases, and low-effort affiliate spam.

# How the script figures out the deals:

1. **Scraping & Ingestion:** Continuously monitors and scrapes incoming deal streams asynchronously.
2. **Price & Glitch Detection:** Parses the text to extract MRP, actual selling price, and discount percentages. It filters out marketing fluff (like "Flat 70% Off" when there’s no real drop) and only flags genuine discount thresholds.
3. **AI Categorization (Gemini):** Sends the clean product context and metadata to Gemini to classify items into strict buckets (Sneakers, Electronics, Gym/Supplements, Beauty, Fashion)
4. **Auto-Dispatching:** Instantly routes the verified deal into its dedicated category channel, so if you only want sneakers or gadgets, you don't get spammed with sarees or phone covers.

# Tech Stack:

* Python 3.11
* SQLite
* Google GenAI SDK
* Deployed on a GCP

You can search @**LootBoxMain** on Telegram if you want to get the offers live 

document fetched 2026-10-14 00:24 · canonical https://reddit.com/r/developersIndia/comments/1x51sti/wrote_a_program_to_detect_ecommerce_price_drops