SubClear

The AI agent that autonomously kills unwanted subscriptions — no phone calls required

Comment

Inspiration

We've all been there: you sign up for a 7-day free trial that takes one click to join, but when you try to cancel, it turns into a nightmare. You have to call during business hours, mail a physical letter, or navigate through five pages of "Are you sure?" guilt trips.

We learned this is called a "Roach Motel" dark pattern—easy to get in, impossible to get out. The average consumer wastes $200/month on subscriptions they don't value because the friction to cancel is just too high. We were inspired to build SubClear to give people a "Nuclear Button" for their unwanted bills.

What it does

SubClear is an automated "Cancellation Agency" in your pocket.

  1. Subscription Audit: It connects to your bank feed to detect recurring charges (Netflix, Gyms, Adobe).
  2. The Kill Switch: You swipe left on any subscription you want gone.
  3. Autonomous Execution:
    • For digital services, our AI Agent logs in and navigates the cancellation flow for you.
    • For "offline" services (like Gyms) that demand physical letters, SubClear automatically generates and mails a legal Notice of Cancellation, acting as your limited Power of Attorney to force them to stop billing.

How we built it

We focused on a clean, high-trust architecture:

  • Frontend: Built with HTML/CSS and Vanilla JS to create a lightweight, responsive dashboard. We prioritized accessibility and clear "Red/Green" status indicators.
  • Backend Logic: We designed the system around Python and Playwright.
  • AI Agents: We prototyped the browser automation using Playwright to handle the DOM interactions (clicking specifically named buttons like "Cancel Membership" or "End Benefits").

Challenges we ran into

  • The "Guilt Trip" UI: Companies often hide the "Confirm Cancellation" button behind confusing text (e.g., "Keep My Benefits" is big and green, while "Cancel" is tiny and grey). Designing an agent that doesn't get tricked was a logic challenge.
  • Legal Compliance: figuring out how to handle "offline" cancellations. We realized we couldn't just "hack" a gym's database; we had to interface with the real world via certified mail.

Accomplishments that we're proud of

  • The "Kill Switch" UI: We managed to distill a complex 45-minute phone call process into a single button press on our dashboard.
  • Full Coverage: We didn't just solve the "easy" cancellations (Netflix); we designed a solution for the "hard" ones (Gyms) too.

What we learned

We learned that "Friction" is architecture. It's not accidental that cancelling is hard. To fight it, we can't just be better negotiators; we need better automation. We also learned how powerful Browser Automation can be for consumer protection—turning the tools usually used for testing into tools for empowerment.

What's next for SubClear

  • LLM Negotiation: Adding a chat-bot that can also negotiate lower rates for bills you want to keep (like WiFi).
  • One-Time Card Integration: partnering with Privacy.com to issue "Burner Cards" so users never have to give their real credit card to a free trial again.

Built With

Share this project:

Updates