A claim has emerged that companies looking to raise competitiveness by adopting AI should prioritise using their ways of working and unique internal knowledge over securing the latest models. The argument is that even if they introduce AI agents, it is hard to achieve the expected results if they do not change existing operating practices.

SiliconANGLE reported on Oct. 20 that John Roese (존 로이스), Dell Technologies' global chief technology officer and chief AI officer, presented this view at the Dell Technologies Symposium in New York.

Dell has used autonomous AI agents in internal work for about 2 years. Based on that, it stressed that companies should go beyond simply adding AI tools and redesign organisational structures, work processes and workforce allocation.

Roese pointed out, "If a company introduces AI without changing how it operates, it may remain at the level of tacking new technology onto existing work." He said the essence of introducing AI agents is to redivide work that people handled by sharing it anew with machines.

Another factor Dell is focusing on is specialised knowledge accumulated inside companies. Dell is organising employees' work experience and judgement criteria into skill files that AI can read.

The files are simple documents in Markdown format, but they contain company-specific work rules and decision criteria that are not disclosed externally.

AI agents use them to write code, calculate deal prices and provide customer service to Dell's standards instead of generating generic answers.

Another feature is that agents accumulate new knowledge and relationships between tasks as they handle work. Dell is linking this into a knowledge graph and using it to develop new digital work capabilities.

Roese stressed, "Since this knowledge is a core asset that differentiates it from competitors' AI, it should not be handed over indiscriminately to outside companies." He also said, "There is no problem in itself with using cutting-edge AI models or open models." He argued that a company's real competitiveness lies not in AI models but in the proprietary knowledge it has accumulated and executed differently from competitors.

Competition in enterprise AI is expanding beyond model performance into operating innovation and independent knowledge management.