Insights

Agentic AI: Elegant Workaround or Enterprise Revolution?

LLMs weren’t originally designed or trained to function as agents in enterprise settings. They weren’t built to take actions, complete structured tasks, or act as digital labor within an organization.

They also have no knowledge — by design, and for good reason — of the internal transactions, processes, or how an organisation delivers its products and services.

They can, however, reason and generate natural language, both extremely well.

So while the emerging frameworks for agentic AI are exciting, the whole pattern still feels, at its core, like an elegant workaround.

That said, it is worth experimenting, evolving, and hoping that a more complete agentic scaffolding will form around LLMs — one that makes AI truly productive in an enterprise context.

All insights