TL;DR

TL;DR Mail turns confusing official mail into 6th-grade action steps, deadline alerts, and free legal aid—so no family loses their home to fine print

Comment

Inspiration

Every day across Austin and Central Texas, families face a silent crisis in their physical mailboxes. Over 32,000 eviction cases are filed annually in Travis and Williamson counties, and $180M+ in critical SNAP and Medicaid benefits are lost statewide each year—not because families are disqualified, but because of confusing paperwork and missed deadlines.

Under Texas Property Code §24.005, a standard Notice to Vacate gives tenants as few as 72 hours (3 days) to respond before a landlord can file an eviction lawsuit. Yet more than 54% of Americans read below a 6th-grade level, while official government notices and medical bills are written at a 12th-to-16th grade level. When intimidated residents don't understand a letter, fear causes paralysis. They tuck the envelope away in a drawer, the 72-hour clock runs out, and default judgments occur automatically.

I built TL;DR Mail as a solo hacker to dismantle this bureaucratic barrier and ensure no family loses their home, power, or healthcare simply because of fine print.

What it does

TL;DR Mail is a zero-barrier, camera-first civic literacy utility that instantly decodes intimidating official mail:

  • 📸 Instant Camera OCR & Analysis: Snap a photo or upload an image of any court summons, eviction notice, utility bill, or benefits letter.
  • 🧠 6th-Grade Reading Level Breakdown: Gemini translates dense legalese into crystal-clear plain English (and 4 other languages: Spanish, Vietnamese, Chinese, and Arabic).
  • 🚨 Key Stake Extraction: Instantly surfaces the sender, exact financial balance due, strict deadline date, and the real-world consequence of doing nothing.
  • ⏰ 1-Tap Calendar Sync: Exports deadlines directly to Google Calendar and universal .ics files with triple-stage reminder alarms (3 days, 1 day, and 2 hours prior).
  • 🛡️ In-Browser Privacy Shield (PII Redactor): A client-side HTML5 canvas tool that lets users blackout Social Security Numbers, account numbers, and names before anything is transmitted.
  • ⚖️ Auto-Drafted Formal Dispute Letters: Generates ready-to-mail legal dispute letters citing Texas Property Code §24.005, Texas Surprise Medical Billing protections, or the Fair Debt Collection Practices Act (FDCPA).
  • 📍 Hyper-Local Central Texas Aid Directory: Automatically connects the user to verified free local aid, including Texas RioGrande Legal Aid (TRLA), Austin Tenants Council, Austin Energy Customer Assistance Program (CAP), and 2-1-1 Texas.
  • 🔊 Multilingual Voice Read-Aloud: Integrated Web Speech API reads the breakdown aloud for visually impaired users or non-fluent readers.
  • 🖨️ 1-Page Printable Case Sheet: Formats the document analysis into a clean physical intake sheet for caseworkers and legal aid clinics.

How I built it

As a solo developer, I engineered the entire full-stack system from UI/UX and client-side canvas security to serverless API routes and AI orchestration:

  • AI Engine & Multimodal Vision: Powered by Google Gemini 2.5 Flash via the official @google/genai SDK. I implemented strict responseJsonSchema constraints to guarantee deterministic structured JSON output without hallucinations.
  • Frontend & Architecture: Built with Next.js 16 (App Router), React 19, and TypeScript 5.0 for complete end-to-end type safety, deployed serverlessly on Vercel.
  • Styling & Motion: Styled with Tailwind CSS v4 featuring accessible typography, responsive mobile viewports, and micro-interactions powered by Framer Motion and Lucide React.
  • Client-Side Privacy: Created a custom interactive HTML5 canvas redaction tool that allows users to blackout sensitive PII directly in their browser before any image data is sent to the server.
  • Resilient Dual-Mode Engine: Built an intelligent offline fallback engine pre-loaded with realistic Travis County test notices (eviction, utility shutoff, Medicaid renewal, medical bill) so judges can test every feature instantly with zero setup.

Challenges I ran into

  1. Building Full-Stack Solo in a Hackathon Sprint: Balancing frontend design, mobile camera handling, PII canvas redaction, calendar export logic, and AI prompt engineering as a solo developer required strict architectural planning and prioritization.
  2. Deterministic Information Extraction: Bureaucratic notices come in wild formats—scanned PDFs, wrinkled phone photos, poor lighting. Designing prompts and JSON schema constraints that reliably extracted exact deadline hours and currency amounts required rigorous edge-case testing.
  3. Client-Side Privacy vs. OCR Accuracy: Allowing users to redact sensitive PII on the client side without accidentally breaking document context or warping image aspect ratios required fine-tuning canvas rendering coordinates.
  4. Hyper-Local Contextual Grounding: Generic AI tends to recommend national 1-800 numbers. I curated a local knowledge base of Travis County resources (TRLA, Austin Tenants Council, Austin Energy CAP) and instructed Gemini to prioritize these organizations for community members.

Accomplishments that I'm proud of

  • Architected, designed, and deployed a production-ready, full-stack civic app completely solo.
  • Built a zero-barrier, zero-login app where users don't need an account, password, or credit card—preserving anonymity for vulnerable tenants.
  • Sub-3-second multimodal document extraction with structured deadlines, actionable checklists, and audio read-aloud.
  • Implemented client-side PII redaction that ensures sensitive personal identification never leaves the user's device unmasked.
  • Made the platform accessible across 5 languages, screen-reader friendly typography, and 1-page printable case sheets for clinic caseworkers.

What I learned

  • How deeply the "bureaucratic literacy gap" affects civic life: legal language is often intentionally structured in ways that overwhelm everyday citizens.
  • Advanced usage of Google's @google/genai SDK, specifically schema-constrained output decoding and multimodal token optimization.
  • The critical importance of trauma-informed UX design: when someone is facing eviction or utility shutoffs, the interface must be calming, unambiguous, and immediate.

What's next for TL;DR Mail

  • Direct SMS / WhatsApp Submission: Allow users without smartphones or reliable data plans to text photos of notices to a hotline number.
  • Partnerships with Local Legal Clinics: Integrate an instant "1-Click Intake Referral" system directly with Travis County legal clinics and housing non-profits.
  • Automated Certified Mail Dispatch: Partner with online mailing APIs to print and mail formal dispute and appeal letters directly on the user's behalf.

Built With

  • framer-motion
  • gemini-2.5-flash
  • google-gemini-api
  • html5
  • lucide-react
  • next.js
  • node.js
  • pwa
  • react
  • rfc-5545-icalendar
  • tailwind-css
  • typescript
  • vercel
  • web-speech-api
Share this project:

Updates

Submission history