DeCipher

Inspiration

My mother, a first-generation Taiwanese immigrant, now naturalized citizen who has lived and worked in America for over two decades, recently faced a frustrating property management dispute. While she had to return to Taiwan for a family emergency around December, her property manager took advantage of her absence, charging unauthorized fees and breeching agreement. Despite her determination and work ethic, my mother's basic English proficiency made it nearly impossible for her to navigate the complex legal language in management agreements or effectively advocate for herself when confronted with questionable charges. She couldn't understand why she was being billed for services she never approved, or why the property manager refused to provide receipts or recognize her authorized representative.

As I helped her decode the legal jargon, draft formal communications, and prepare for mediation, I realized this wasn't just my mother's problem. Millions of hardworking Americans and non native English speakers face similar situations daily - vulnerable to exploitation simply because they can't fully understand the legal language used against them.

DeCipher emerged from this deeply personal challenge. I wanted to create a tool that could have empowered my mother to independently understand her rental agreement, identify unauthorized charges, and confidently assert her rights - breaking down the language barriers that too often lead to financial exploitation and stress.

What it does

DeCipher is like having a patient friend who knows law inside and out. No judgment, no complex jargon, just clear, helpful explanations when you need them.

Translates legal documents into simple, accessible language
Provides multi-language support for non-native English speakers
Offers patient, judgment-free assistance available 24/7
Analyzes pros and cons specific to your unique situation
Delivers factually accurate information you can trust Our advanced AI breaks down complex legal documents into simple, accurate explanations that everyone can understand:
Extracts text from any document using advanced OCR technology
Supports PDFs, photos, scanned documents, and even handwritten notes
Allows users to highlight specific phrases or sections for targeted explanation
Enables direct conversation with documents while preserving the exact legal language

How I built it

Bolt.new - Primary development platform for the application Grok 3 & Grok 3 Mini - Used for debugging and refining the logical flow of the application Perplexity Pro/Domain Specific Filter - Conducted research on legal use cases and implemented a state law verification module to ensure accuracy Agentic Document Extraction - OCR functionalities, text extraction and object descriptions

Challenges we ran into

As a non-programmer, the technical aspects were challenging. The most significant hurdle was optimizing text processing while maintaining verbatim legal language, which is critical for accuracy in legal contexts. Implementing effective chat history context and structuring outputs to be both legally precise and user-friendly required multiple iterations. Balancing simplification with legal accuracy proved particularly difficult. Right after my last Hackathon I quickly discover I am on a very short time frame to get this done, with only 14 days I hope I can make it in time!

Accomplishments that I am proud of

While AI made this project possible, I'm particularly proud of my prompt engineering and system design. The application successfully helped resolve my mother's property management dispute by organizing all relevant documents, communications, and legal information in one place. Creating a unified system that could analyze the complete timeline of events (from August 2024 through May 2025) and identify specific contract violations made a real-world difference in a complex situation. Perplexity has helped us tremendously throughout this process with real relevant results and fact checking for reliable sources was invaluable.

What I learned

I learned to start with thorough planning before implementation and to keep designs simple but comprehensive. The importance of understanding sentiment hierarchy and user intent became clear - especially when users may have language barriers or limited subject knowledge. I discovered that AI responses must be carefully guided to address the user's actual needs rather than just answering surface-level questions.

What's next for DeCipher

Enhanced chat history & contextual understanding to better follow complex legal situations
Optimization for efficiency without sacrificing factual accuracy
Fine-tuned smaller models for sentiment evaluation to better understand user emotions
Expanded voice support and multi-language capabilities to serve diverse communities
Development of specialized modules for common legal situations (housing, employment, immigration)

Thanks for the opportunity - demo site offer free plan so everyone can try but does require authentication.