A new survey of 500 senior IT leaders reveals that AI success correlates more with integration maturity than model intelligence. While 76 percent of companies report having at least one AI workflow in production, most are struggling to move beyond isolated pilots.
The study from MIT Technology Review Insights, in partnership with automation platform Celigo, shows 90 percent of organizations with at least one AI workflow fully in production already utilize an integration platform, and 39 percent with enterprise-wide integration have deployed AI across multiple departments.
In stark contrast, just one percent or less of organizations without an integration platform have been able to scale AI beyond a single department.
“Enterprises are quickly realizing that AI strategy is really architecture strategy,” says Ronen Vengosh, chief strategy officer at Celigo. “The problem is that most business processes span multiple systems, but if you’re using AI tools within your CRM or ERP, for example, the information comes from within those applications. If AI cannot see across systems, it cannot reason across them.”
As AI systems become increasingly autonomous, the report highlights integration platforms as a foundational layer that enables consistent data access, cross-system orchestration and governance. Without this layer, organizations remain trapped in silos. 59 percent of enterprise-wide integration platform users draw from five or more data sources for their AI workflows and no organizations without an integration platform were able to reach the same level of data complexity, limiting their AI to basic, single-source tasks.
Beyond initial deployment, the report shows that leaders are moving into the next phase of maturity with mostly autonomous workflows. In these high-growth environments, the integration platform has transitioned into a vital governance layer that ensures AI remains reliable and resilient to changing business needs. 34 percent of organizations with enterprise-wide integration have already achieved mostly autonomous AI workflows, while 95 percent of executives say their AI workflows already have some level of autonomy, with most expecting autonomy to increase over the next 12 to 18 months.
“We are seeing a clear shift from AI experimentation to a focus on the infrastructure required for scale,” says Jan Arendtsz, founder and CEO of Celigo. “This report proves that to move toward agentic workflows that drive actual value, companies must bridge the gap between their disparate applications and their AI models.”
You can get the full report from the Celigo site.
Image credit: BiancoBlue/depositphotos.com
