Subscriptos

Block sneaky subscriptions. We find the ToS traps and generate burner cards to keep your wallet safe.

Comment

Inspiration

In an era where digital convenience often masks predatory billing, we recognized the immense potential of AI to level the playing field between corporations and consumers. Subscriptions and free trials have become "Information Asymmetry" traps, with hidden aggressive auto-renewal clauses and data-sharing agreements within tens of pages of complex legalese, knowing that users won't read them. Inspired by the need for financial transparency, we envisioned Subscriptos: a proactive AI-driven platform that translates legalese into understandable metrics. Our goal is to empower users with the tools to understand exactly what they are signing and provide a secondary layer of defense through burner payment methods, regardless of the platform.

What it does

Subscriptos is a comprehensive security suite that delivers real-time financial protection through intelligent analytics and secure virtual payment infrastructure:

  • Integrated Chrome Extension: Automatically detects subscription-based websites and offers instant ToS scanning before you commit.
  • AI-Driven Safety Scoring: Powered by Google’s Gemini LLM, the system generates a robust risk profile across three vectors: Data Privacy, Integrity, and Consumer Fairness, providing scores (0-100) and detailed justifications.
  • Stripe-Powered Virtual Card Issuing: Allows users to generate virtual cards on the fly for subscriptions and free trials. Users can enjoy the trial without the fear of hidden charges or the hassle of manual cancellations.
  • Intelligent Dashboard: A centralized "Command Center" that synthesizes all user activity, current virtual card statuses, and historical scans.
  • Gen Z "Brainrot" Translation: An AI-optimized mode that translates complex legal contracts into modern slang for a quick, intuitive vibe check.
  • Interactive AI Legal Assistant: An embedded chatbot allowing users to ask follow-up questions about specific clauses, selling of data, or cancellation policies using the Gemini-powered chat module.
  • Multi-Modal Uploads: Support for analyzing terms via URL detection, plain-text copy-pasting, or direct PDF/File uploads.

How we built it

We combined React, Vite, FastAPI, Python, Gemini, Supabase, PostgreSQL, Stripe, and Postman.

  • Frontend (React + Vite): Built with React 19 and Vite for lightning-fast performance and an optimized developer experience.

  • Styling & UI: Leveraged Tailwind CSS and Framer Motion for sleek, premium micro-animations.

  • Icons & Assets: Integrated Lucide React for a clean icon system and customized credit card asset visualizations.

  • Backend (FastAPI + Python): Developed a robust, asynchronous API using FastAPI to handle complex requests between the frontend, LLMs, and databases.

  • PDF Parsing: Engineered a custom document ingestion pipeline using PyMuPDF to clean and process raw PDF data for AI consumption.

  • Authentication: Integrated AuthContext patterns to manage user sessions and secure API access.

  • AI & Machine Learning: Utilized Google Gemini 2.5 Flash for high-speed, accurate inference.

  • Prompt Engineering: Developed sophisticated system instructions for structured JSON output, ensuring consistent scoring and multi-modal Gen Z translations.

  • Interactive Inference: Enabled real-time follow-up assessments through Gemini’s session-based chat pipelines.

  • Data & Security Infrastructure: Employed Supabase (PostgreSQL) to maintain a persistent history of scans and card data and a way for users to authenticate to our service.

  • Payment Infrastructure: Integrated Stripe's Issuing API to provide secure, real-time virtual card generation.

  • Development Workflow: Utilized Postman for comprehensive API verification testing and PowerShell for automated extension packaging scripts.

Challenges we ran into

  • We initially struggled with session management and endpoint verification using Postman, as our Supabase JWTs frequently expired.
  • As our project became more complex and coupled between the frontend and backend, merging branches occasionally caused localized environment breakages. We overcame this by standardizing our .env configurations and ensuring strict version control practices.

Accomplishments that we're proud of

We are so proud of successfully building and linking a complex multi-service application in an incredibly shortened timeframe, completing core functionality in the first half of the hackathon, and look forward to getting 8 hours of sleep before judging starts.

What we learned

  • We learned how to build and route via FastAPI, creating independent modules for LLM inference, Stripe integration, and database operations that our frontend could call seamlessly.
  • Integrating Stripe's Issuing API was a massive learning curve, teaching us the intricacies of virtual card generation and secure data handling.
  • Each of us experimented with an unfamiliar technology and learned how to implement it into a full-stack project.

What's next for Subscriptos

  • Our ultimate goal is to evolve from passive "observation" to active "management." By linking legitimate bank accounts, Subscriptos will soon be able to directly identify, track, and officially cancel subscriptions on behalf of the user directly from the dashboard.
  • We plan to officially publish the extension to the Google Chrome Web Store to eliminate the need for manual developer-mode installations.
  • Implementing a comprehensive native Dark Mode for the web app and extension to improve accessibility and aesthetics.

Built With

Share this project:

Updates