PayPilot — AI Agentic Checkout

• 

  thumbnail

• 

  paypilot1

• 

  paypilot2

• 

  paypilot3

PayPilot — AI Agentic Checkout

Online checkout is still a maze of search filters, comparison tabs, and form fields. PayPilot replaces it with a conversation: tell the AI what you want, and it handles product discovery, selection, and the complete PayPal checkout for you.

How it works:

1. You describe what you're shopping for in natural language — e.g. "wireless headphones under $100".
2. PayPilot's agent parses your intent (product keywords + budget constraints) and ranks matching products from the catalog.
3. You pick an item and confirm. PayPilot calls the PayPal Sandbox Orders API (v2) to create a real order with intent=CAPTURE.
4. You approve the payment on PayPal's hosted approval page.
5. You type "done" — PayPilot captures the payment and shows your receipt.

PayPal integration (central to the project):

• POST /v2/checkout/orders — order creation with purchase units, amounts, and application context
• POST /v2/checkout/orders/{id}/capture — payment capture after buyer approval
• Runs against the PayPal Sandbox — no real money moves during the demo.

AI: The PayPilotAgent (backend/agent.py) is a state-machine conversational agent: intent parsing, product matching, and checkout orchestration (search, pick, confirm, order, approve, capture). It is designed model-agnostic — any LLM can replace the rule-based intent parser via the llm_backend hook without changing the PayPal flow.

What's next:

• Plug in a production LLM for richer intent understanding
• Expand the catalog via real merchant APIs
• Multi-currency and PayPal Vault for returning customers

Built With

• flask
• html/css
• javascript
• paypal
• python
• rest