Only 11% of organizations have an agent in production. The firm advocates for a redesign of processes and agent management modeled on that of a workforce.
Deloitte's finding in his Tech Trends 2026 is severe: the gap between the interest in AI and its production remains wide. 30 per cent of organizations are at the exploration stage, 38 per cent are at the pilot's stage, but only 14 per cent are ready to deploy and 11 per cent actually have an agent in production. At the same time, 42% are still developing their strategy and 35% have no formal plan. Gartner anticipates the failure of 40% of AI projects by 2027.
Three infrastructure obstacles
The firm first identifies legacy systems, which do not offer real-time execution or modern APIs. Next, the data architecture: ETL-powered warehouses do not provide the accessible context that an officer needs, and nearly half of organizations are struggling on research (48%) and re-use (47%) of their data. Finally, governance, largely absent when a system decides alone.
The central recommendation has one formula: not to be content with "bitumer the cow's way". Rather than automating chains designed for human operators, it is necessary to map the value chain and redraw flows from what agents can do. Deloitte proposes to treat the latter as a "silicon workforce": integration, performance monitoring by immutable logs, life cycle management, zero-confidence authentication. Employees would migrate to compliance, governance and innovation functions.
Read the full article on the original source
Go to the source →