DiscountHub — AI Shopping Platform

DiscountHub is a live shopping platform with real-time deals, promo codes, mobile apps, and AI-powered discovery—connecting shoppers and AI agents through WebMCP and a universal MCP server.

Comment

DiscountHub — AI Shopping Platform

Inspiration

Every day, online stores publish thousands of discounts, special offers, and promo codes. But finding a genuinely useful deal is surprisingly difficult.

Shopping information is fragmented across marketplaces, affiliate networks, and individual retailers. Prices change, promotions expire, and search results often include products without meaningful discounts.

We built DiscountHub to address this problem: a unified platform that helps people discover real shopping opportunities without searching through dozens of stores.

Our vision goes beyond a traditional deals website. We want shopping information to be accessible wherever users need it — on the web, on mobile devices, and through AI assistants.

What it does

DiscountHub is a live, cross-platform shopping discovery ecosystem.

It combines a public website, mobile applications, a FastAPI backend, automated data pipelines, and AI integrations into one connected platform.

Real deals and promo codes

DiscountHub aggregates discounted products, promotions, and promo codes from connected retail and affiliate sources, including eBay and participating Awin advertisers.

Its data processing system applies strict validation rules to help users discover meaningful offers rather than ordinary product listings.

The platform checks discount information, prices, product availability signals, images, and freshness. Outdated offers are automatically excluded from public results based on source-specific freshness policies.

Web and mobile experience

Users can discover and explore offers through:

  • The public DiscountHub website.
  • Cross-platform mobile applications developed with Flutter.
  • Search and filtering features for finding relevant discounts.
  • Store links that take shoppers to the original merchant.

DiscountHub is a discovery platform, not a checkout marketplace. Purchases take place directly with the relevant retailer.

AI-powered shopping discovery

DiscountHub also makes its shopping data accessible to AI agents.

The platform supports WebMCP for browser-based agent interactions and an independent Model Context Protocol (MCP) server for compatible external AI clients.

Through structured tools, AI assistants can search for discounted products, discover available promotions, explore store offers, and identify promising shopping opportunities.

This creates a reusable shopping intelligence layer that can serve multiple AI experiences, including Alexa+.

Instead of building an isolated demo for a single assistant, we are extending an already operational platform with an interoperable interface for AI-powered commerce discovery.

How we built it

DiscountHub uses a multi-platform architecture:

Frontend: A public website and Flutter applications for mobile platforms.

Backend: Python and FastAPI, providing the central API and business logic.

Data infrastructure: SQLite-backed storage, automated provider synchronization, product normalization, duplicate handling, and freshness-aware public queries.

Retail integrations: eBay Browse API and Awin product and promotional data sources.

AI interoperability: WebMCP and an independent MCP server using Streamable HTTP, exposing structured shopping discovery tools to compatible AI agents.

A major architectural goal was to keep the existing website, mobile experience, backend, and APIs operational while introducing new AI capabilities.

Challenges we faced

Separating real discounts from ordinary products

Not every marketplace listing with a price is a discount. We developed validation rules that require meaningful pricing information and filter out unsuitable listings.

Keeping offers current

Large retail catalogs change continuously. We implemented automated synchronization, source-specific freshness windows, expiration checks, and cleanup mechanisms to reduce stale results.

Working across different commerce data sources

Retailers and affiliate platforms provide inconsistent fields and formats. Our backend normalizes these sources into a common model that can be used by the website, applications, API, and AI tools.

Making the platform useful to AI agents

AI integrations need more than unstructured product pages. We designed MCP tools that expose discoverable, structured shopping information while preserving the existing platform and its validation rules.

Accomplishments

We are proud that DiscountHub is not just a hackathon mockup.

It is an operational project with a public website, mobile applications, real retail integrations, a backend, automated synchronization, and AI-accessible shopping tools.

The same core infrastructure supports both human shoppers and AI agents.

We have also focused on continuous improvements to search relevance, data quality, provider coverage, and operational reliability rather than simply increasing the number of listings.

What we learned

Building a useful shopping discovery platform requires more than collecting products.

Data freshness, discount credibility, search relevance, compatibility, and reliable integrations are essential.

We also learned that MCP can extend an established application into the broader AI ecosystem without replacing the user experience people already have.

What's next

We plan to expand the number of integrated retailers, improve deal discovery and ranking, increase the coverage of quality-checked offers, and develop richer AI-assisted shopping experiences.

Our long-term goal is to make DiscountHub a dependable shopping discovery infrastructure for people and AI assistants across multiple platforms.

Built With

Share this project:

Updates

Submission history