Meet the bunq Purchase Advisor

Most banking tools help after money is spent. We asked: what if bunq could help users make smarter decisions before they buy? Impulse purchases, bad deals, and overlooked alternatives cost people money every day. We wanted to turn banking from expense tracking into purchase intelligence.

bunq Purchase Advisor helps users answer one simple question: Should I buy this?

How does the Purchase advisor work?

Users can scan a product, paste a URL, or enter a product manually. Our multimodal AI analyzes the item, compares alternatives and vendor pricing, considers spending context, and returns a clear recommendation:

BUY — good value, aligned with budget

WAIT — poor timing or likely better later

CHOOSE ALTERNATIVE — similar value at lower cost

Instead of only tracking spending after purchases happen, bunq can help optimize spending before it happens.

How we built it

We built the prototype with:

Next.js + TypeScript

React + Tailwind CSS

Anthropic multimodal AI

Product recommendation logic engine

Mock transaction/spending context integration

System flow:

Image / URL / Text Input → Product Analysis → Alternatives + Price Insight → Financial Recommendation

Biggest challenges:

  • Keeping recommendations simple enough, but meaningful
  • Differentiating “advisor” from generic shopping comparison tool
  • Making AI feel trustworthy for financial decisions
  • Designing something that feels natural to the bunq product experience

Main Lesson: Less complexity made the product stronger.

What's next for bunq Purchase Advisor?

Potential next steps:

  • Real-time bunq account personalization
  • Live merchant/vendor integrations
  • Price history and “buy later” predictions
  • Personalized financial goals-aware recommendations
  • Agentic shopping assistant inside bunq

Long term, we see this as moving from expense management to financial decision intelligence.

Built With

  • anthropic
  • bunq-inspired
  • multimodal-ai
  • next.js
  • react
  • recommendation-engine
  • spending-intelligence
  • tailwind
  • typescript
  • vercel
Share this project:

Updates