AI is becoming cheaper, yet enterprise AI bills may rise as agents perform more work. The real challenge is measuring returns, controlling consumption and governing increasingly autonomous AI systems.

Agentic AI is shifting from armies of specialist agents toward capable general agents that load skills, tools, knowledge and workflows on demand, using subagents only where independent agency adds value.

Generative AI has transformed what computers can understand and create, but many of the hardest enterprise AI problems remain familiar. As AI moves from answering questions to taking actions, the important question is becoming not simply how intelligent AI is, but how much authority we should give it.