-
-
deal engine. running in progress 01
-
deal engine. running in progress 02
-
deal engine. running in progress 03
-
deal engine. running in progress 04
-
deal engine. result 01
Inspiration
Online merchants often lose customers at checkout because of price differences, even when a better offer could help close the sale.
I wanted to explore how AI could turn price hesitation into completed purchases, while balancing price transparency and merchant profitability.
What it does
Deal Engine is an AI-powered checkout workflow that helps merchants make competitive, margin-protected offers.
When a customer checks prices, Deal Engine retrieves live competitor offers and uses two AI agents to recommend whether to match the lowest price, offer a partial discount, or keep the original price.
Customers can accept the offer or proceed at the original price through PayPal checkout.
How it was built
I built Deal Engine using Next.js, TypeScript, PostgreSQL, Supabase, and Vercel, integrating three core technologies:
- Channel3: Provides product information and live competitor offers.
- Meta Muse Spark: Powers two AI agents.
- PayPal: Enables Sandbox checkout with server-authorized transaction amounts.
A deterministic validation layer ensures AI proposals comply with merchant pricing policies.
Challenges
My biggest challenge was making AI pricing decisions commercially meaningful rather than simply matching the lowest price.
I also had to distinguish real market observations from simulated merchant data, handle incomplete competitor information, and prevent AI recommendations from bypassing merchant pricing rules or payment authorization.
Accomplishments
I built a working multi-agent checkout workflow connecting live competitor pricing, simulated merchant business intelligence, AI-assisted negotiation, and PayPal Sandbox payments.
Most importantly, merchants remain in control: AI recommends prices, but only policy-compliant offers can proceed to checkout.
Take Aways
I learned that AI can go beyond recommendations to support commercial decisions.
Combining specialized agents with market intelligence creates more context-aware pricing opportunities, while deterministic safeguards keep financial transactions controlled.
Built With
- channel3
- javascript
- meta-ai
- next.js
- node.js
- paypal
- postgresql
- react
- supabase
- tailwind
- typescript
- vercel
