According to Gartner's latest forecast, by 2028, 80% of generative AI commercial applications will be developed on existing data management platforms. This shift is expected to reduce development complexity and cut delivery time in half.
Currently, the development of generative AI commercial applications mainly relies on combining large language models (LLMs) with enterprise internal data, as well as evolving technologies such as vector search, metadata management, prompt design, and embedding technology. However, without a unified management approach, enterprises may adopt "dispersed technologies," leading to extended delivery times and increased costs.
Image source note: The image was generated by AI, and the image authorization service provider is MidJourney.
Gartner emphasized the importance of retrieval-augmented generation (RAG) in developing generative AI applications at the recent Data and Analytics Summit held in Mumbai, India. RAG is a framework that enhances the accuracy and reliability of generative AI models and is becoming the foundation for deploying generative AI applications. Gartner pointed out that RAG can provide "flexible implementation methods, enhanced explainability, and the ability to combine with LLMs."
Prasad Pore, senior analyst at Gartner, stated that RAG helps improve processes and automate tasks in multiple business functions, such as sales, human resources, IT, and data management. Currently, data engineers and data professionals face many challenges when developing, testing, deploying, and maintaining complex data pipelines and applications. Pore noted that traditional data management processes are time-consuming and require a lot of manual work, while the application of RAG can significantly increase productivity and simplify the data governance process.
In addition, Pore mentioned that generative models like LLMs themselves are static and work only based on their trained data, lacking the latest information. Through RAG, companies can incorporate the latest business or organization-specific data into the models to enhance the effectiveness of generative AI applications in answering questions, analyzing logs, and making decisions.
When talking about the types of generative AI commercial applications, Pore said they can be divided into three main categories: process improvement and automation (such as corporate knowledge management and document processing automation), user experience (such as customer support automation and personalized shopping experiences), and insights and predictions (such as conversational business intelligence and data discovery).
When building and deploying generative AI applications, Gartner suggests that enterprises consider the following points: first, evaluate whether existing data management platforms can be transformed into RAG-as-a-service platforms; second, prioritize RAG integration and incorporate vector search, graph, and chunking technologies from existing data management systems; finally, utilize metadata and operational data to protect intellectual property, address privacy issues, and prevent malicious use.
Key Takeaways:
🌟 By 2028, it is expected that 80% of generative AI commercial applications will be developed on existing data management platforms, cutting delivery time in half.
🚀 Retrieval-Augmented Generation (RAG) will become an important foundation for developing generative AI applications, providing flexibility and explainability.
🔍 Gartner recommends that enterprises assess the transformation potential of existing platforms, integrate RAG technology, and use metadata to protect security.