ASEAN enterprises train LLMs with enterprise data, study finds

ASEAN enterprises are now leading in data intelligence, training their large language models (LLMs) with enterprise data.

This is the findings of a new economist impact report by data, analytics, and artificial intelligence company Databricks.

The study, titled "Unlocking Enterprise AI: Opportunities and Strategies", examines the challenges businesses face in adopting and scaling AI, and the techniques they are using to drive greater value from these investments.

As finance leaders navigate the ever-changing landscape of digitalisation, it is important to take note of how enterprises in the region move for better budget planning and to drive growth.

Although the report found that nearly half of data scientists across all of APAC still rely on general-purpose LLMs without contextual enterprise data, it should be noted that these general-purpose models often lack the quality, governance, and evaluative capabilities that enterprise-specific data can bring. 

"It's clear that AI is becoming an integral part of every business, and the technology is emerging as a critical driver of business growth. Yet enterprises remain cautious, balancing ambition with concerns around quality, cost, and implementation," says Cecily Ng, vice president and general manager for ASEAN at Databricks.

"ASEAN organisations need AI platforms that prioritise data privacy, centralise governance, and deliver a sustainable total cost of ownership (TCO) at scale."

Additional key findings of the report include: 

• The vast majority of ASEAN enterprises (91%) are using GenAI in at least one function. However, only one in three (32%) believe their GenAI applications are production-ready. Respondents across the Asia Pacific cite key hurdles, including cost (40%), skills (38%), governance (38%) and quality (33%). 
• Only 18% of ASEAN respondents believe AI is overhyped. In fact, 77% see the technology as crucial to their long-term goals. Despite the momentum, 37% believe investment across technical and non-technical domains is insufficient.
• By 2027, 99% of all ASEAN respondents expect GenAI adoption across both internal and external use cases.  
• ASEAN organisations expect to mix and match different models and tools in their Agent Systems, spanning open source and proprietary technologies, to drive better performance. By 2027, 94% plan to deploy open source AI models. 
• Just 18% of ASEAN respondents are confident their organization can secure enough AI talent. 
• Only 20% of ASEAN respondents strongly agree that their organization’s data and AI governance are sufficient.Enterprises face challenges with fragmented data estates, complicating discovery, access permissions, data usage, audits and sharing. Governing AI models and tools is also vital to meet evolving AI regulations. To succeed, enterprises need a unified and open governance approach.