Data unification and AI Engine

 The Equitus.us Knowledge Graph Neural Network (KGNN) platform works like a backend data unification and AI engine optimized to run natively on IBM Power servers (specifically Power10 and Power11), which benefits IBM Power users by providing a high-performance, edge-ready, and secure environment for AI workloads.

________________________________________________________________________

KGNN Functionality and IBM Power Integration

The KGNN is not a traditional "API" in the sense of a simple REST endpoint for a single task, but rather a data platform that makes AI-ready, semantically-rich data accessible to various applications, which can include APIs for downstream use.

1. Automated Data Unification:

   • What it does: KGNN automatically ingests, cleans, structures, and links disparate structured and unstructured data sources (PDFs, logs, databases) into a self-constructing knowledge graph.² This is often referred to as "Auto ETL" (Extract, Transform, Load) and semantic mapping.³

   • IBM Power Benefit: By running natively on IBM Power10/11, the process leverages the server's architecture, including the Matrix Math Accelerator (MMA) built into the Power processor cores.⁴ This allows for high-performance data processing and AI inferencing without the need for external GPUs, reducing cost and complexity.⁵

2. Semantic Contextualization:

   • What it does: It transforms siloed data into an interconnected, schema-less knowledge graph, automatically identifying and linking entities, relationships, and context.⁶

   • IBM Power Benefit: The native optimization on Power ensures this complex, real-time contextualization can be performed efficiently, even at the edge or in disconnected, secure environments, which are common use cases for IBM Power systems.⁷

3. AI-Ready Data and Querying:

   • What it does: The knowledge graph is vectorized (converted into numerical representations) and indexed, making it ready for AI applications like Retrieval-Augmented Generation (RAG), Large Language Models (LLMs), and traditional Machine Learning (ML).⁸ It supports federated queries and bidirectional integration with other applications.⁹

   • IBM Power Benefit: The architecture is designed to make this AI-ready data available efficiently across applications running on the Power platform. The security features of Power10, such as transparent memory encryption, help maintain data security and control for sensitive enterprise and government data.¹⁰

---

In summary, for IBM Power users, KGNN acts as a powerful, Power-native middleware that rapidly transforms raw enterprise data into a usable, intelligent knowledge graph, eliminating manual data prep and maximizing the built-in AI acceleration capabilities of the IBM Power hardware.¹¹

The following video is an introductory discussion on running real-time analytics on IBM Power with Equitus.

How to Run Enterprise AI Without GPUs: IBM Power + Equitus Real-Time Analytics