ReguPilot

Architecture design

Due to connection issues with our EC2 servers, we’ve relaunched the project at:

https://main.d3nq32obaeyqaw.amplifyapp.com/

Inspiration

As AI continues to evolve rapidly, compliance is no longer optional — it's essential.
We noticed that early-stage AI teams often lack the legal resources and technical bandwidth to assess privacy, security, and regulatory risks before launch. We wanted to change that.

What it does

ReguPilot is an AI-powered compliance copilot designed to help AI developers, PMs, and startups evaluate legal, privacy, and security risks before launch.

Users can upload project files or describe their AI product in plain language. With a single click, ReguPilot analyzes the input using a multi-agent system and generates a structured compliance report — including risk scores, regulatory flags, triggered clauses, and actionable recommendations.

How we built it

Frontend: Next.js, TypeScript, Tailwind CSS, and Chart.js
Backend: Flask, AWS Bedrock, PyMuPDF, boto3, RAG with Bedrock Knowledge Base
Agents:
- Agent 1: File parsing and project understanding
- Agent 2: Content screening
- Agent 3: Policy matching via RAG and a legal knowledge base
- Agent 4: Risk scoring and summary generation

Challenges we ran into

As first-time builders on AWS, we faced a steep learning curve.
It took considerable time to familiarize ourselves with various services.

Accomplishments that we're proud of

Successfully built and deployed a functional multi-agent compliance review system within 10 days.
Integrated AWS Bedrock and RAG pipelines into a meaningful legal evaluation workflow
Designed a clean, explainable user experience from upload to report.