Inspiration

Modern companies face massive GDPR fines because they often lose track of personally identifiable information hidden inside complex and undocumented databases. We wanted to solve the Right-to-be-Forgotten problem with a fully autonomous AI system.

What it does

Oblivion Protocol is an autonomous multi-agent compliance engine powered by Vertex AI and MongoDB MCP.

When a deletion request is received, the Schema Crawler Agent maps the database from scratch and discovers live PII vectors. The Compliance Decision Agent uses Gemini to semantically classify the discovered fields and determine whether data should be permanently deleted or anonymized. Finally, the Execution Agent performs the wipe and generates a tamper-evident legal receipt containing the AI's reasoning.

The system transforms complex data compliance into an automated and auditable process.

How we built it

We built the frontend using React, Next.js, and TypeScript.

For intelligence and reasoning, we used Google Vertex AI with Gemini. MongoDB and the official MongoDB Model Context Protocol server were used for database access and schema discovery. The system follows a multi-agent architecture consisting of the Schema Crawler Agent, Compliance Decision Agent, and Execution Agent.

Challenges we ran into

One of the biggest challenges was building a system that could understand database schemas without relying on hardcoded rules. Another challenge was preserving analytics data while still ensuring legal compliance. Designing the semantic reasoning layer and generating verifiable audit receipts required careful coordination between multiple agents.

Accomplishments that we're proud of

We successfully built a complete autonomous compliance pipeline capable of mapping, classifying, deleting, and auditing personal data. We also implemented a semantic reasoning layer that enables intelligent deletion decisions instead of simple rule-based approaches.

What we learned

This project taught us how powerful agentic AI systems can be when combined with semantic reasoning and structured database access. We also learned the importance of explainability and auditability in enterprise AI systems.