FreelancePay AI

Freelancers earn globally but get paid painfully. FreelancePay AI invoices clients via PayPal, chases late payments, flags risky clients, and shows what you actually keep.

Comment

Inspiration

India has millions of freelancers (developers, designers, writers) who work for clients abroad, and most of those clients pay through PayPal. But getting paid is the painful part:

  • Invoices are created by hand in spreadsheets or Word.
  • Late payments from overseas clients take constant chasing.
  • Fees and currency conversion quietly eat into the income.
  • There's little protection when a client disputes or disappears.
  • Most freelancers couldn't tell you what they earned last quarter or what they actually keep.

We wanted one tool that handles the whole cycle: invoice, get paid, understand your money.

What it does

FreelancePay AI is an AI finance copilot for global freelancers, built on the PayPal developer platform.

  • Smart invoicing: type a sentence like "Invoice Acme $800 for the logo project, due in 14 days" and the AI creates a PayPal invoice in USD, EUR, GBP and more.
  • Payment tracking and reminders: see every invoice's status (draft, sent, paid, overdue) and approve AI-written reminders in a polite, firm or final-notice tone.
  • AI copilot: ask "Who owes me money?", "What did I earn this quarter?" or "Which client pays slowest?" and get answers from your real data, plus a 30-60 day cash-flow forecast.
  • Fee and conversion insights: see gross amount, PayPal fee, conversion loss and net amount for each payment, with tips to keep more.
  • Client risk flags: each client gets a reliability score based on payment speed, overdue history and disputes, so you can think twice before taking on a risky one.
  • Milestone escrow: a client authorizes payment, the funds are held, and they're captured when the client approves the work. Refunds are handled through PayPal if there's a dispute.
  • Spending and savings assistant: ask "Can I afford a $150 laptop accessory?" and the AI checks your income, upcoming expenses and buffer.
  • Subscription manager: tracks tool subscriptions and flags ones you rarely use, with estimated yearly savings.
  • Tax set-aside estimator: suggests how much of each payment to put aside (an estimate, not tax advice).

Safety by design: the AI never acts alone. It proposes actions such as sending a reminder or issuing a refund, and the user approves each one.

How we built it

PayPal integration (central to the app)

  • Invoicing API: create, send and track invoices.
  • Orders API (authorize and capture): the milestone escrow flow.
  • Refunds API: dispute handling.
  • Transaction Search: payment history and fee data.
  • Webhooks: real-time updates when an invoice is paid or a refund happens.

AI layer

  • An LLM ([Claude / Gemini / OpenAI, fill in what you used]) with tool calling. The model calls functions such as list_invoices, get_client_summary, draft_reminder, forecast_cashflow and calculate_fees.
  • All numbers (totals, fees, forecasts, risk scores) are computed in plain code, and the AI explains and acts on them. This keeps the figures accurate instead of letting the model guess.

Stack

  • Frontend: React + TypeScript, Tailwind, Recharts
  • Backend: [FastAPI / Node.js] with [SQLite / PostgreSQL]
  • Hosting: Render
  • Payments: PayPal Sandbox

Challenges we ran into

  • [Example: handling PayPal webhook events reliably and keeping invoice status in sync.]
  • [Example: designing the authorize-then-capture escrow flow so funds are held correctly.]
  • [Example: getting the AI to give accurate answers by moving all calculations out of the prompt and into code.]
  • Sandbox has no real transaction history, so we built realistic seeded data to test the copilot and forecasts.

Accomplishments that we're proud of

  • A complete invoice-to-payment flow running end to end on the PayPal sandbox.
  • An AI copilot that answers from real data and always asks for approval before acting.
  • [Add anything specific you achieved.]

What we learned

  • [Example: how PayPal's invoicing, order and webhook systems fit together.]
  • [Example: why tool calling with deterministic code beats asking an LLM to do the math.]
  • [Add your own.]

What's next

  • Support for more currencies and payout methods popular with Indian freelancers.
  • Region-specific tax guidance.
  • Automatic client onboarding and contract templates.
  • Mobile app with payment notifications.

Notes for judges

  • This project was newly created during the hackathon submission period.
  • All payments use the PayPal Sandbox, and demo data is seeded to show realistic scenarios.
  • Fee and tax figures are estimates, not financial or legal advice.
  • Setup and run instructions, plus test credentials, are in the README of our repository.

Built with

PayPal Invoicing API, PayPal Orders API, PayPal Webhooks, React, TypeScript, [FastAPI/Node.js], [database], [AI model], Render

Built With

Share this project:

Updates

Submission history