Verbis Graph - GraphRAG Retrieval Engine | Knowledge Graph Traversal

A dedicated GraphRAG container on AWS Marketplace, purpose-built for Amazon Bedrock Agents. Verbis Graph Engine enables organizations to deploy a secure, scalable knowledge retrieval layer directly within their own AWS environment - supporting customer-controlled deployment, data residency, and private AWS environments.

By combining vector similarity search with knowledge graph traversal, Verbis Graph Engine captures relationships between entities across documents, enabling multi-hop reasoning, explainable responses, and citation-backed outputs designed to improve retrieval quality compared with vector-only RAG systems.

It is designed for enterprise AI applications requiring AI governance,

Supports AI governance, AI compliance, AI explainability, and audit-ready documentation workflows.

Combines vector search with knowledge graph traversal to deliver context-aware, relationship-aware results beyond traditional vector-only retrieval. Enables multi-hop reasoning across entities and documents.

All responses are linked to exact source documents, providing full traceability and auditability - critical for compliance and regulated environments.

Runs inside the customer AWS environment via Amazon ECS. Amazon ECS, with Amazon EKS support available for Kubernetes-based deployments, ensuring:

• Full data residency
• VPC-level isolation
• Alignment with enterprise security policies

Supports BYOC (Bring Your Own Cloud) and BYOK (Bring Your Own Key) for complete control over infrastructure, data residency, and encryption.

Compatible with LangChain, LlamaIndex, AutoGen, CrewAI, and Amazon Bedrock Agents for seamless integration into existing AI pipelines.

Designed for low-latency retrieval and scalable workloads in production environments.

Traditional RAG systems rely on vector similarity, which may miss relationships across documents.

Example: When asking "Which marketing campaigns were affected by the supply chain disruption in Q3?" - vector search retrieves similar documents but cannot connect related entities. GraphRAG models relationships across documents, enabling multi-hop reasoning and more complete, context-aware answers.

This matters most in regulated environments where accuracy, traceability, and explainability are non-negotiable.

1. Deploy the container in your AWS environment via ECS or EKS
2. Ingest documents - knowledge graph and vector embeddings build automatically
3. Query via REST API or connect to your preferred AI framework
4. Retrieve answers with citations and full reasoning paths

Watch the full setup walkthrough:  https://www.youtube.com/watch?v=JhXqYwpJHlE 

• Subscribe to the product in AWS Marketplace
• Deploy using Amazon ECS or Amazon EKS
• Ensure port 8080 is open in your security group

• Obtain the public IP or load balancer URL after deployment
• Navigate to the application on port 8080

• Upload a document for indexing
• Wait for the indexing process to complete
• Ask questions in plain language
• Explore the generated knowledge graph visualization

• Swagger UI available at your endpoint under /docs

Amazon Bedrock - Amazon Neptune - LangChain - LlamaIndex - AutoGen - CrewAI - OpenAI - Anthropic Claude

Verbis Graph Engine is deployed entirely within the customer AWS environment, supporting:

• Data residency and control
• VPC-level isolation
• Integration with IAM and enterprise security policies

The platform follows AWS Well-Architected best practices for deployment.

A 1-month pilot and Proof of Concept period is available on request for evaluation in real-world use cases. Contact support@verbisgraph.com  to request access.

Paid services available for custom architecture and deployments.

Built by Prodigy AI Solutions - Enterprise support available on request.