Organizations worldwide are set to build the vast majority of their generative AI business applications on existing data management platforms rather than starting from scratch, according to new research from Gartner Inc.

The technology research firm predicts that by 2028, 80% of generative AI business applications will be developed on organizations' existing data management infrastructure. This approach is expected to reduce both the complexity and time required to deliver these applications by 50%.

The forecast comes as companies rush to integrate AI capabilities into their operations but struggle with the technical challenges of connecting large language models to their internal data systems.

"Building GenAI business applications today involves integrating large language models with an organization's internal data and adopting rapidly evolving technologies like vector search, metadata management, prompt design and embedding," said Prasad Pore, Senior Director Analyst at Gartner.

"However, without a unified management approach, adopting these scattered technologies leads to longer delivery times and potential sunk costs for organizations," Pore added.

The Rise of RAG Technology

Central to this shift is retrieval-augmented generation (RAG), which Gartner identifies as a cornerstone technology for AI application deployment. RAG enhances the accuracy of large language models by combining them with business-specific datasets, addressing a key limitation of current AI systems.

"Most LLMs are trained on publicly available data and are not highly effective on their own at solving specific business challenges," Pore explained. "However, when these LLMs are combined with business-owned datasets using the RAG architectural pattern, their accuracy is significantly enhanced."

The technology works by retrieving relevant information from a company's data repositories to provide context for AI responses, making the systems more accurate and explainable for business use cases.

Strategic Recommendations

To capitalize on this trend, Gartner recommends that enterprises evaluate whether their current data management platforms can be transformed into RAG-as-a-service platforms. The research firm also advises companies to prioritize integrating RAG technologies such as vector search and graph databases from existing solutions rather than building standalone systems.

Additionally, organizations should leverage both technical and operational metadata to protect their AI applications from security risks, privacy issues, and intellectual property leaks.