A recent survey from K2view, the generative data product company, examines the top challenges enterprises are facing as they work toward generative AI (GenAI) implementation. The report unveils the most significant roadblocks to realizing GenAI’s full potential lie in organizations’ existing data infrastructure, particularly in the areas of data accessibility and latency, data privacy, and security.
The responses came from 300 senior professionals who are directly involved in the planning, building, or delivery of GenAI applications.
Let's drill in to understand the details.
Assessing GenAI adoption in enterprises
GenAI adoption is accelerating as enterprises recognize its transformative potential. Organizations are strategically moving from experimentation to pilot projects, focusing on areas like marketing, sales, and customer operations where GenAI can deliver substantial benefits. Techniques like Retrieval Augmented Generation (RAG) are becoming essential tools for customizing AI models to meet specific business requirements.
The RAG revolution
While generic Large Language Models (LLMs) offer impressive capabilities, they often fall short of meeting specific business needs without customization. The survey shows that only 14% of organizations utilize off-the-shelf LLMs in their AI projects. The vast majority – 86% – are enhancing their LLMs using techniques like Retrieval Augmented Generation (RAG), fine-tuning, and embedding.
RAG is rapidly gaining traction as a leading method for enhancing GenAI models with specific organizational data, with 60% of respondents currently piloting RAG. Industry-specific adoption rates provide further insights.
These figures demonstrate strong momentum in adopting RAG, particularly in sectors where data privacy and precise responses are critical. However, the limited number of organizations that have transitioned to full production reflects the complexities involved in scaling these advanced technologies effectively.
Data is critical to realizing GenAI’s full potential
Enterprise data presents one of the biggest challenges for GenAI deployment, and organizations are still struggling with it as they move their GenAI projects to production.
48% of respondents cite data security and privacy concerns, and 33% cite enterprise data readiness as roadblocks to deployment. The difficulty lies in the fragmented nature of enterprise data, which is often spread across multiple systems and analytical data stores, making it hard to integrate, govern, and make accessible in near-real time, under stringent guardrails, to GenAI applications.
Growing focus on leveraging the technology for customer operations
According to our survey, Marketing and sales (63%) followed by Customer operations (54%) are the primary areas where companies plan to implement generative AI solutions in the next 12 months.
While marketing and sales have some of the strongest and most mature use cases for GenAI, there’s a growing focus on leveraging the technology for customer operations across various industries.
Customer service departments across industries are now deploying GenAI-powered solutions that transform traditional support models. These implementations are yielding measurable improvements in key performance metrics – from faster resolution times to enhanced personalization of customer interactions.
Download your copy here
Top comments (0)