VendorClause AI | Legal Document Intelligence Platform

An end-to-end AI document intelligence platform for legal-style workflows with async ingestion, hybrid retrieval, and streaming source-grounded answers.

PolyGot (VendorClause-ai) is a multi-service AI platform for legal and compliance document workflows. It provides secure auth and session management, PDF upload with asynchronous ingestion, hybrid retrieval across vector and graph stores, and source-grounded streaming chat answers. The monorepo is organized into client, Gateway, and ai-service modules, with Docker Compose orchestration for local infrastructure such as Postgres, Redis, MinIO, Qdrant, Neo4j, MongoDB, Ollama, and Langfuse.

Tech Stack

Next.js 16React 19TypeScriptZustandSocket.IONode.jsExpressPrismaBullMQRedisMinIOPythonLangGraphLangChainLlamaParseQdrantNeo4jDocker Compose

Features

Secure authentication and session management through the Gateway service.
PDF upload with async ingestion pipeline using queues and worker processing.
Hybrid retrieval pipeline combining Qdrant vector search and Neo4j graph traversal.
Streaming, source-grounded chat answers delivered via Redis pub/sub and Socket.IO.
Dedicated MCP servers for Qdrant and Neo4j tool access in retrieval workflows.
LangGraph-based retrieval workflow with supervisor, rewriting, orchestration, fusion, CRAG evaluation, and generation stages.
End-to-end local environment with Docker Compose and service-level observability.
Demo video support embedded directly from repository docs.

Challenges

Coordinating ingestion and retrieval across multiple services while preserving reliable async state transitions.
Designing low-latency streaming UX from query workers through Redis channels and Gateway Socket.IO bridges.
Balancing vector and graph retrieval quality for legal-style, source-sensitive answers.
Managing local developer experience across many infra dependencies and service processes.

Learnings

Improved system design skills for AI-first, event-driven multi-service applications.
Deepened hands-on understanding of retrieval orchestration with LangGraph and MCP servers.
Strengthened deployment and debugging practices for complex local stacks using Docker Compose.
Built stronger patterns for source-grounded answer generation and stream-based product UX.

Link

LiveGithub

This app is not live at the moment.

VendorClause AI | Legal Document Intelligence Platform

An end-to-end AI document intelligence platform for legal-style workflows with async ingestion, hybrid retrieval, and streaming source-grounded answers.

Features

Secure authentication and session management through the Gateway service.

PDF upload with async ingestion pipeline using queues and worker processing.

Hybrid retrieval pipeline combining Qdrant vector search and Neo4j graph traversal.

Streaming, source-grounded chat answers delivered via Redis pub/sub and Socket.IO.

Dedicated MCP servers for Qdrant and Neo4j tool access in retrieval workflows.

LangGraph-based retrieval workflow with supervisor, rewriting, orchestration, fusion, CRAG evaluation, and generation stages.

End-to-end local environment with Docker Compose and service-level observability.

Demo video support embedded directly from repository docs.

Challenges

Coordinating ingestion and retrieval across multiple services while preserving reliable async state transitions.

Designing low-latency streaming UX from query workers through Redis channels and Gateway Socket.IO bridges.

Balancing vector and graph retrieval quality for legal-style, source-sensitive answers.

Managing local developer experience across many infra dependencies and service processes.

Learnings

Improved system design skills for AI-first, event-driven multi-service applications.

Deepened hands-on understanding of retrieval orchestration with LangGraph and MCP servers.

Strengthened deployment and debugging practices for complex local stacks using Docker Compose.

Built stronger patterns for source-grounded answer generation and stream-based product UX.