VEDA - Verified Enterprise Decisions & Auditing

Making MSME compliance faster smarter and trustworthy, with AI powered document intelligence, plus rule based checks , and explainable audit automation.

Comment

Inspiration

MSMEs often struggle with compliance because business information is spread across multiple documents and verification is largely manual. This leads to delays, errors, and missed opportunities. We wanted to build an AI-powered platform that simplifies compliance while keeping every decision transparent and evidence based.

What it does

VEDA (Verified Enterprise Decisions & Auditing) helps businesses upload documents, extract key information using OCR, verify it against user provided data, evaluate compliance through deterministic rules, generate AI powered explanations, visualize relationships with a knowledge graph, and produce actionable compliance plans with audit-ready evidence.

How we built it

VEDA is built using React, TypeScript, Vite, and Tailwind CSS on the frontend, with FastAPI, Python, SQLAlchemy, and Alembic on the backend. It uses PostgreSQL for structured data, Neo4j for knowledge graphs, Mistral OCR for document extraction, Groq (Llama 3.1) for explainable AI, Supabase for authentication and storage, and Docker for containerized deployment. We also implemented end-to-end Playwright testing to verify the complete workflow.

Challenges we ran into

Our biggest challenges were integrating OCR with structured business data, preventing unreliable AI outputs by combining deterministic rules with LLM explanations, building a persistent knowledge graph, and ensuring every compliance decision was traceable through comprehensive end to end testing.

Accomplishments that we're proud of

-Built an end-to-end AI-powered compliance platform for MSMEs, from document upload to explainable compliance decisions. -Combined OCR, deterministic rule-based reasoning, AI explanations, Neo4j knowledge graphs, and action planning into a single workflow. -Successfully integrated React, FastAPI, PostgreSQL, Neo4j, Supabase, Groq, and Mistral OCR using Docker. -Verified the complete application with automated end to end Playwright tests covering the entire user journey.

What we learned

We learned that trustworthy AI requires more than an LLM. Combining rule-based reasoning, structured knowledge graphs, explainable AI, and rigorous testing results in a far more reliable enterprise solution

What's next for VEDA

We plan to integrate official government verification APIs where available, expand compliance coverage, add multilingual support, and provide proactive compliance reminders and analytics for businesses and auditors.

Built With

Share this project:

Updates