Inspiration
Compliance is a critical issue in the tech industry that affects companies of all sizes, from small-to-medium enterprises to industry giants. This is repeatedly underscored by the massive fines levied for regulatory non-compliance. As developers, our passion is building innovative products, but we repeatedly collide with the harsh reality of bureaucratic red tape required to avoid exorbitant penalties.
That friction was the core inspiration for the Compliance Intelligence Platform (CIP). We wanted to remove compliance as a barrier to innovation by building a solution to seamlessly and automatically map data and handle compliance regulations, allowing engineering teams to focus on what they do best—building.
What it does
CIP acts as an automated, localized compliance officer. It securely connects to a company's internal data sources (such as emails, file servers, and code repositories) to map out exactly how sensitive data flows through the organization. Using this knowledge graph, the platform leverages AI to automatically generate, track, and score required compliance documents (like the EU AI Act Risk Assessment or GDPR Privacy Policies) without requiring massive manual audits.
How we built it
The core architecture is decoupled and secure by design:
- Data Collection: We utilized an OpenClaw agent tasked with regularly scanning a simulated company environment (including regular files, emails, and a GitHub integration). As it discovers assets and transfers, it pushes structured nodes and edges to our platform via a REST API.
- The Central Brain: The backend is built with Spring Boot 4 and Kotlin, storing the incoming telemetry in PostgreSQL to form an adjacency list data flow graph. It automatically scans ingested text to apply sensitivity flags for PII (personal identifiable information), secrets, and financial data.
- Compliance Intelligence Engine: To guarantee data privacy, the platform's core intelligence uses a Local LLM (Ollama). This ensures sensitive company data never leaves the internal network while still allowing the system to intelligently analyze data flows.
- Interactive Frontend: We built the user interface with Next.js and shadcn/ui, featuring a real-time, interactive canvas that visualizes the corporate data graph so compliance officers can easily trace where sensitive data travels.
Challenges we ran into
During the hackathon, the biggest hurdles we encountered involved real-world data collection and going live.
Specifically, data crawling and polling emails proved far more complex than anticipated when trying to consistently map disparate unstructured data into our clean compliance graph without losing context.
Additionally, we faced significant demo deployment challenges. While CIP is designed to be an entirely on-premise application (and we successfully tested it running locally on a Mac Studio leveraging Ollama), migrating the heavily integrated system to Hetzner and switching out the local models for OpenAI to ensure reliable remote access for the judges required rapid infrastructure pivots at the eleventh hour.
Accomplishments that we're proud of
We are incredibly proud of successfully integrating the OpenClaw agent to autonomously make sense of chaotic, unstructured data (like raw emails) and deterministically map it into a sleek, explorable visual graph. Furthermore, designing an architecture capable of running entirely on-premise to guarantee absolute data privacy while still delivering powerful AI-driven insights is a major milestone for our team.
What we learned
This project pushed us to deeply explore the mechanics of autonomous data collection. We significantly deepened our understanding of OpenClaw's interaction models and setup. We learned how to effectively instruct and stabilize an external agent to autonomously process complex, messy corporate environments and successfully map them into deterministic API payloads for our backend.
What's next for CIP - Compliance Intelligence Platform
First and foremost, we want to deploy CIP in a real-life context by dogfooding it on our own active projects. By running it against our own repositories, servers, and communication channels, we can battle-test its capabilities and ensure it truly removes friction for developers. Beyond that, we plan to expand our data connectors to natively integrate with more enterprise tools like Jira, Microsoft Teams, and Confluence. We also want to implement real-time alerting for active compliance breaches (e.g., instantly notifying the security team if a developer commits a database password to GitHub or sends PII to unauthorized external domains). Finally, we aim to further refine our document generation engine to support even stricter international compliance frameworks (like HIPAA or SOC2).
Built With
- kotlin
- next.js
- ollama
- openai
- pgvector
- postgresql
- react
- remotion
- springboot
- typescript
Log in or sign up for Devpost to join the conversation.