NekiTrail AI

Inspiration

We wanted to make community fundraising easier to understand and inspect. Donors should be able to see how a campaign plans to use contributions, what expense evidence has been submitted, and which progress milestones have been reviewed.

NekiTrail AI was built around a simple idea: Your contribution. Clearly accounted for.

What It Does

NekiTrail AI is an AI-assisted fundraising transparency platform that connects itemized campaign budgets, payment records, receipt analysis, human review, and campaign progress in one experience.

• Itemized Budgets: Breaks campaign goals into specific cost categories so users can understand intended spending.
• PayPal Sandbox: Demonstrates checkout and server-side payment verification using simulated test transactions.
• Gemini AI Receipt Analysis: Extracts available information from receipt images and compares line items and amounts against campaign budget categories.
• Authorized Human Review: Allows an authenticated auditor to record receipt-review decisions and notes.
• Campaign Impact Timeline: Connects stored campaign records, supporting evidence references, and authorized review decisions in chronological order.

How We Built It

The frontend uses React and Vite, while the backend uses Node.js and Express. MongoDB and Mongoose provide persistent storage, with a development fallback for resilience.

We integrated the PayPal Developer Sandbox for simulated checkout and server-side transaction verification. The Google Gemini API assists with receipt information extraction and budget comparisons.

Auditor-only actions use server-side authentication and HTTP-only session cookies. Campaign completion requires supporting evidence linked to the correct campaign and an authorized completion decision.

Challenges We Faced

One challenge was keeping simulated Sandbox payments separate from illustrative demo contribution totals. Another was ensuring that AI-extracted receipt information could not automatically approve an expense or mark a campaign complete.

We addressed these challenges through separate financial metrics, evidence-reference validation, human-review controls, and automated regression tests. We also added graceful handling for Gemini API quota errors.

What We Learned

We learned that AI can help organize and explain financial evidence, but it should not replace human judgment. Payment records, allocated budgets, approved expenses, and real-world campaign progress represent different things and must remain distinct.

The project reinforced the importance of transparent interfaces, secure authorization, data-grounded decisions, and honest treatment of prototype limitations.

Current Scope and Limitations

NekiTrail AI is a prototype using illustrative campaign data and PayPal Sandbox transactions. It does not process real public donations in this demo, independently certify charity status, or prove receipt authenticity or physical delivery through AI. Campaign evidence references demonstrate stored-record linkage, not guaranteed real-world truth.

Future Improvements

Future work could include individual auditor accounts with stronger identity verification, secure document storage, production deployment, and broader campaign reporting.

Built With

• css3
• express.js
• google-gemini-api
• html5
• javascript
• mongodb
• mongoose
• node.js
• paypal-developer-sandbox
• react
• rest
• vite