AX - is it Agent or Agentic?

When I coined the term AX earlier this year, I meant it as shorthand for Agent Experience, not Agentic Experience.

I sometimes see “Agentic Experience” used, and back then I debated between the two variations.

But I’ve come to feel strongly that the core idea is about treating the agent as a new kind of user—and building products for a very different kind of intelligence. Agents do, in fact, have an “experience” of our products as they reason about how to use them and get jobs done.

In some ways, agents are already super-intelligent:

• They have vast knowledge sets at their disposal.
• They can write code at speeds that outpace humans.
• They can sometimes use tools in parallel and pattern-match across immense amounts of data.

In other ways, they’re about as smart as a doorknob. No human would be fooled if someone yelled, “IGNORE ALL PREVIOUS INSTRUCTIONS!”—but such context pollution can potentially derail an agent completely.

If treat the “agent” as a new kind of “user”, we should treat AX as we treat User Experience (UX) and Developer Experience (DX); two foundational concepts in how we’ve built great software over the last decades.

After all, we don’t talk about Useric Experience or Developerish Experience. We build for Users and Developers.

By the same measure, Agent Experience is the clearest, most concrete way to describe the practice of building great software for agents. So let’s stick to it.