LegalPulse AI

Inspiration

Everyday freelancers, small business owners, and individuals sign complex legal agreements every day—from IT service contracts to leases and NDAs—without fully understanding the legal jargon or hidden financial liabilities. Hiring a lawyer for every review is expensive and slow. We built LegalPulse AI to bridge this legal literacy gap, providing an instant, accessible legal co-pilot that breaks down agreements and protects people from predatory terms.

What it does

LegalPulse AI is an end-to-end legal contract parser and workflow agent that transforms passive contract review into an actionable, state-specific safety check:

• 🏛️ State Law Audit Engine: Validates clauses against specific US state statutes (e.g., California law) using real-time search verification.
• 🔴 Interactive Legal Review & Analysis: Highlights risky clauses directly on screen with plain-language explanations and guidance.
• 📄 OCR Document Ingestion: Processes contract images and PDFs seamlessly, extracting text for state-level analysis.
• 💬 Ask-Your-Contract Chat: A conversational agent with PostgreSQL chat memory that answers questions with direct references to state law.

How I built it

We built LegalPulse AI using a robust multi-agent architecture:

• Automation & Multi-Agent Orchestration: Custom n8n workflows handling complex agent routing, document parsing, and external searches.
• AI Models & Search: OpenAI Chat Models paired with Tavily Search API for up-to-date US statutory law verification.
• Database & Session Storage: PostgreSQL database maintaining persistent chat history and document memory.
• Frontend App: Built with Next.js and Tailwind CSS for a fast, responsive dark-mode dashboard.

Challenges I ran into

Handling messy OCR outputs from scanned document images and coordinating multi-step state compliance checks without losing context were key technical hurdles. We solved this by creating a structured DB pipeline where full text is extracted, stored in PostgreSQL, and queried dynamically during chat sessions.

Accomplishments that I'm proud of

• Successfully orchestrating end-to-end multi-agent execution using n8n and PostgreSQL memory.
• Deploying a live, working production interface on a custom domain with fast state-specific legal responses.

What I learned

Building an agentic workflow for legal tech requires strict grounding to avoid hallucinations. Enforcing strict statutory context and real-time legal web searches builds deep user trust.

What's next for LegalPulse AI

We plan to expand state-level compliance coverage to all 50 US states, integrate electronic signature workflows, and offer automated compliance monitoring for recurring contracts.

Built With

• google-gemini
• n8n
• next.js
• postgresql
• supabase
• tailwind-css
• tavily-search-api