Melissa Kwok

AI agents assert relationships such as compatible products, cleared counterparties, or approved suppliers, derived from LLM reasoning rather than verified data. When these assertions are wrong, downstream business decisions fail.

GraphRAG combines knowledge graphs with RAG to produce explainable responses grounded in semantic relationships between concepts, entities, and underlying content. Agentic GraphRAG takes this further, allowing agents to autonomously decide which GraphRAG tools to use to achieve information goals.

Graph-based agentic reasoning empowers AI systems to dynamically interact with the rich context and relationships in an Amazon Neptune graph, enabling intelligent, flexible, and scalable AI solutions that extract valuable insights from complex, interconnected data.

Retrieval Augmented Generation (RAG) applications use the power of generative AI to analyze private datasets. GraphRAG combines knowledge graphs with RAG to produce explainable responses that are grounded in the semantic relationships between concepts, entities, and the underlying content.

In this chalk talk, explore how to build a scalable serverless hybrid search application that combines vectors and graph data. Learn how this approach empowers you to leverage the benefits of vector-based search with the rich relationships defined in graphs to provide more comprehensive and explainable answers.

Retrieval Augmented Generation (RAG) applications use the power of generative AI to analyze private, differentiating datasets. However, baseline RAG can sometimes produce poor responses that lack explainability and contextual awareness and include conflated sources and spurious claims.

In this chalk talk, explore how to build a scalable serverless hybrid search application that combines vectors and graph data. Learn how this approach empowers you to leverage the benefits of vector-based search with the rich relationships defined in graphs to provide more comprehensive and explainable answers.

Fraud happens. However, fraud is usually not a random set of acts undertaken by a lone bad actor; rather, it is usually a well-orchestrated, coordinated, and layered approach to defeating systems.

A security graph represents assets that you want to protect and the relationships between these assets and roles within your organization, such as computer networks, telecommunications networks, distribution routes, payment systems, and other parts of your IT infrastructure. In this builders’ session, model these assets, use relationships in your data to consider how different dimensions of...