Singapore – Local-based developers are rapidly adopting artificial intelligence tools, but fragmented infrastructure and unclear enterprise strategies are preventing many organisations from fully scaling their AI initiatives, according to new research from Twilio.
The survey found that 96% of respondents already use AI tools in their daily workflows. However, the report suggested that widespread adoption has exposed operational issues tied to disconnected systems, poor integration, and inconsistent governance.
According to the findings, 46% of respondents identified constant context-switching between disconnected tools as the main source of workflow friction. Meanwhile, 35% said they struggle with incompatible tools, while 24% cited siloed data across multiple systems as a major obstacle.
The survey also pointed to a lack of strategic direction from company leadership. Fewer than 30% of respondents said their organisations have a clear AI deployment strategy, while 41% of founders and startup leaders reported they are still experimenting with AI without a formal adoption framework.
The report found that organisations without a structured AI roadmap were significantly more likely to face difficulties moving projects into production. Nearly 31% of companies lacking a formal AI approach struggled to operationalise initiatives, compared to just 3% among those with a defined strategy.
Twilio’s findings also highlighted differing priorities between technical and product teams. While 61% of software engineers ranked API availability as a key consideration when evaluating tools, only 36% of product managers viewed interoperability as equally important.
The survey suggested that these differing priorities can contribute to fragmented technology decisions, resulting in disconnected systems and expanding data silos over time.
The research also showed that organisations are beginning to move beyond basic AI implementations. Nearly 40% of respondents said they are developing autonomous AI agents, while 25% are integrating Voice AI technologies to manage more complex workflows through natural language interactions.
According to the report, these emerging systems are designed to perform tasks such as scheduling meetings and processing customer refunds across enterprise platforms. However, the survey warned that fragmented infrastructure increases operational risk in such environments, potentially leading to execution failures and compounded errors.
“Running next generation models on fragmented legacy architecture is becoming a liability in today’s agentic ecosystem,” said Michelle Duke, Senior Developer Evangelist at Twilio. “The missing link is the connective tissue between these isolated systems. Unlocking real AI productivity now requires a foundational infrastructure layer to integrate every endpoint.”
