#Build04 · Quick commerce · AI agents

QCom Spy Agents

AI agents that track competitor pricing and availability by pin code, delivered as a daily email report.

additional revenue

higher availability

tracking

Problem

D2C brands had no visibility into what competitors were doing on quick-commerce — pricing, stock, and category positioning across Blinkit, Amazon, and social chatter had to be checked manually, if at all.

Solution

The brand owner submits a form — product website link, product name, category. An AI agent scrapes Reddit, Amazon, Blinkit, and other platform scrapers for competitor data, transforms and stores it, then Claude Haiku formats a report which is emailed straight to the owner via Resend.

Process

Built on Gumloop. Started with the intake form and scraping layer, then added transformation and storage once raw data was reliable, and closed the loop with an automated report generation + email step so no manual pull was needed.

Technical details

How it's built.

What runs where, what it talks to, and what was handed over.

Stack

Gumloop (build platform)Supabase (database)Claude Haiku (report generation)Resend (email delivery)

Integrations

Reddit scraperAmazon scraperBlinkit scraperResend for email delivery

Delivered

Intake form (product link, name, category)
Multi-platform competitor scraping agent
Data transformation + storage
Claude Haiku–generated report
Automated email delivery to brand owner

Architecture

How the pieces connect.

Brand owner submits a form → scraping agent pulls competitor data from Reddit, Amazon, and Blinkit → data is transformed and stored in Supabase → Claude Haiku formats a report → Resend emails it to the owner.