Inspiration
I got the idea for IntentGuard from the growing use of AI agents in online shopping. AI agents can help users find products and make purchases, but there are also risks like fake offers, hidden charges, changed prices, and malicious instructions.
I wanted to create a project that helps users stay safe and keeps them in control even when AI agents handle shopping tasks.
What it does
IntentGuard is a trust and security layer for AI-based shopping and payments.
It helps users check their original purchase intent, examine merchant offers, identify suspicious content, and verify transaction details before approval.
The main features include Intent Passport, Merchant Offer Intelligence, AI Threat Protection, Transaction Verification, Human Consent, Audit & Receipt, and Attack Simulator.
It also includes a PayPal Sandbox payment demonstration using test transactions.
How we built it
I developed IntentGuard using React, TypeScript, Vite, and Tailwind CSS for the frontend. I used Python, FastAPI, and SQLite for the backend.
I divided the application into different sections for checking purchase intent, reviewing merchant offers, detecting threats, verifying transactions, and getting user approval.
I also worked on backend APIs, testing, and the PayPal Sandbox demonstration to bring the different parts together.
Challenges we ran into
One of the biggest challenges I faced was connecting the frontend and backend and making sure all the features worked properly.
I also faced some issues with Python packages, frontend build tools, and transaction verification. Finding and fixing these issues took time.
Another challenge was designing a system that could identify suspicious activities while keeping the user involved before a transaction.
Accomplishments that we're proud of
I am happy that I was able to turn my idea into a working project by developing it individually.
I am especially proud of bringing features like Intent Passport, Merchant Offer Intelligence, AI Threat Protection, and Transaction Verification together in one application.
I also completed 160 backend tests successfully and got the frontend type checking and production build working.
What we learned
While building IntentGuard, I learned more about API development, frontend and backend integration, database management, testing, and application security.
I also understood how important it is to verify information and get user approval before allowing AI agents to perform actions involving money.
Most importantly, this project improved my problem-solving skills and gave me experience in building a complete application on my own.
What's next for IntentGuard: The Trust Layer for Agentic Commerce
In the future, I would like to improve the threat detection system and make it more accurate.
I also plan to add support for more languages, test more shopping scenarios, and improve the payment workflow.
My goal is to develop IntentGuard further so that people can use AI agents for online shopping with more confidence while still having control over their money and decisions.
Built With
- css
- fastapi
- html
- javascript
- paypal
- python
- react
- sqlite
- tailwind
- typescript
- vite
