Year: 2024
Type: article
Abstract: This research proposes a Next-Generation Data Architecture (NGDA) for AI-Driven Enterprise SaaS Platforms to address the limitations of fragmented data pipelines, isolated AI services, synchronization delays, and inadequate governance in modern enterprise environments. The proposed architecture integrates heterogeneous enterprise data sources, batch and streaming ingestion, unified lakehouse storage, continuous data-quality monitoring, metadata and lineage management, governance controls, adaptive AI processing, and enterprise SaaS applications into a single end-to-end data ecosystem. The architecture is designed for mixed enterprise SaaS workloads involving transactional processing, analytics, machine learning, generative AI, retrieval-augmented generation (RAG), and real-time event processing. Experimental evaluation compares NGDA with an existing AI-native SaaS architecture using latency, throughput, synchronization delay, infrastructure overhead, data quality, governance effectiveness, AI accuracy, and RAG performance as the primary evaluation dimensions. The experimental results indicate that NGDA substantially improves end-to-end enterprise data and AI performance. Median latency decreases from 428 ms to 286 ms, P95 latency from 812 ms to 521 ms, and synchronization delay from 4.8 s to 1.7 s, while throughput increases from 7,850 to 11,420 requests/s. Overall data quality improves from 91.2% to 97.8%, and lineage completeness increases from 82.4% to 97.6%. AI accuracy improves from 89.4% to 94.7%, while RAG retrieval relevance increases from 84.9% to 93.1% and RAG response accuracy from 85.4% to 93.7%. The unauthorized-context rate is reduced from 2.8% to 0.4%. These findings indicate that integrating data management, governance, and AI processing within a unified architecture can simultaneously improve operational efficiency, data trustworthiness, and intelligent application quality. The proposed NGDA therefore provides a scalable architectural foundation for next-generation enterprise SaaS platforms.
Author: Ravindrakumar Prajapati
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7216820222