Taiwan Companies Face Three AI Agent Adoption Hurdles, Including Data Silos and High Token Costs

Generative AI is evolving from a question-and-answer tool into AI agents capable of making autonomous decisions, connecting to APIs and executing tasks. Infobip said that despite widespread SaaS adoption among Taiwanese companies, deployment remains constrained by fragmented CRM and ERP data, Token processing costs and a gap in technical understanding between front-end and back-end teams. Successful adoption depends on the ability to coordinate across systems.

Infobip’s head of business in Taiwan, Wang Yao-wei, shared the observations following MWC 2026, with the report published on April 18, 2026. The company uses AgentOS to create a framework for data acquisition, solution assessment, task assignment and action execution. Croatian digital insurer LAQO already uses AI agents to handle 30% of customer inquiries, but Infobip did not disclose the project’s cost or Token rates.