BUYWISE — Shop Smarter. Buy with Confidence.
Inspiration
Online shopping provides consumers with thousands of choices, but choosing the right product is not always easy. Shoppers often struggle to understand technical specifications, compare different products, identify hidden limitations, and determine whether a product is actually worth its price.
This problem is particularly relevant to students and young professionals purchasing laptops and other technology products with limited budgets. A product may look attractive in advertisements but have limitations such as non-upgradeable memory, inadequate storage, or specifications that do not match the buyer's actual needs.
We created BUYWISE to address this problem by making product research more intelligent, transparent, and consumer-focused.
What it does
BUYWISE is an AI-powered shopping assistant that helps consumers move from product discovery to informed purchasing decisions.
Its main capabilities include:
- Conversational AI Shopping: Users describe their requirements in natural language, and the AI assistant helps interpret their needs, budget, and intended usage.
- Product Discovery: Users explore available products through search and filtering.
- Product Comparison: A comparison interface presents technical specifications and important differences between products.
- Deterministic Recommendation Engine: Products are evaluated using four weighted factors: specification match (35%), budget (25%), quality (20%), and value (20%).
- Purchase Regret Check: Identifies potential hardware limitations, budget concerns, and unnecessary specifications before purchase.
- PayPal Checkout Architecture: Implements PayPal Orders v2 integration architecture with server-side price validation for eligible BUYWISE demo-store products. The current demonstration uses simulated checkout mode.
- After-Purchase Assistant: Provides return guidance, deadline information, warranty details, policy explanations, and simulated demo-store return requests.
How we built it
BUYWISE was developed using a modern full-stack architecture.
Frontend
- Next.js
- React
- TypeScript
- Tailwind CSS
Backend and Database
- Supabase PostgreSQL
- Supabase Authentication
- Row Level Security (RLS)
- Server-side API routes
Artificial Intelligence
- Google Gemini API for conversational shopping assistance, requirement extraction, product comparison explanations, and post-purchase policy guidance.
Payment Architecture
- PayPal REST API Orders v2 integration architecture.
- Server-side price validation and order processing.
- Webhook verification and idempotency handling.
Development and Testing
- Google Antigravity for AI-assisted development.
- Playwright for end-to-end testing.
- Git and GitHub for version control and source-code publication.
The product discovery layer includes a provider adapter designed for Channel3 integration. The current demonstration uses a seeded demo catalogue because live Channel3 credentials are not configured.
What makes BUYWISE different?
Traditional shopping platforms primarily help users find products. BUYWISE focuses on helping users understand whether a product is actually suitable for them.
Its deterministic recommendation engine separates scoring from AI-generated explanations. This helps ensure that product recommendations follow defined evaluation criteria rather than relying entirely on generated text.
The Purchase Regret Check adds another layer of consumer protection by identifying possible hardware traps and unnecessary spending before the purchase decision.
BUYWISE also extends assistance beyond the purchase through its After-Purchase Assistant.
Challenges we faced
Developing BUYWISE involved several technical challenges:
- Designing a structured product data model capable of supporting consistent product comparisons.
- Connecting natural-language AI interactions with structured shopping requirements.
- Separating AI-generated explanations from deterministic recommendation calculations.
- Implementing server-side price validation and secure payment-processing architecture.
- Designing clear boundaries between external retailer products and eligible BUYWISE demo-store products.
- Managing return policies and warranty information without presenting simulated retailer actions as genuine transactions.
- Testing application workflows, database security policies, and API integrations.
These challenges helped us understand the importance of data reliability, backend security, payment-state verification, and transparent AI behaviour.
What we learned
Through BUYWISE, we learned how to integrate conversational AI into a full-stack application, build deterministic recommendation systems, structure product comparison workflows, design payment-processing architecture, and develop post-purchase support features.
We also learned that AI should assist users with explanations and decision-making without replacing reliable application logic or authoritative payment verification.
Future improvements
Our planned improvements include completing and verifying the PayPal Sandbox transaction flow, enabling live product-provider integration, deploying the application publicly, and expanding product coverage and consumer-support capabilities.
Our vision
We believe shopping technology should not simply encourage consumers to buy more products. It should help them buy the right products.
BUYWISE — Shop smarter. Buy with confidence.
Built With
- artificial-intelligence
- e-commerce
- generative-ai
- google-gemini
- next.js
- paypal
- paypal-orders-api
- playwright
- postgresql
- react
- recommendation-systems
- supabase
- tailwind-css
- typescript