Issue Agent washing and agentic workslop make some deployments net-negative

Summary

Deloitte warns that part of the disappointment with agentic AI stems from misapplied or mislabelled projects. Many initiatives described as agentic are really ordinary automation use cases, and enterprises often deploy agents where simpler tools would do, which leads to poor returns. Vendors add to the confusion by rebranding existing automation products as agents, a practice the report calls agent washing. Poorly designed agentic applications can even add work to a process rather than remove it. Some enterprises have found that this agentic workslop leaves processes less efficient than before. The report's remedy is to examine processes end to end and redesign them around what agents do well, instead of paving the existing path.