Nobody handed Anthropic a trophy at HumanX this week. They didn't need one.

At the Moscone Center in San Francisco, where thousands of AI industry people gathered to talk about agents, automation, and what's next, one name kept coming up in conversations that had nothing to do with Anthropic's booth or panels: It was Claude, not as a flashy announcement. It was just the centre of conversation because people kept mentioning it.

One vendor on the convention floor put it plainly: his team had moved to Claude. OpenAI, he said, had "fallen off." His words, not a press release.

That read is spreading. OpenAI is still enormous; its recent $122 billion funding round and approaching IPO aren't nothing. But the company has had a rough few months for perception.

Last month, it shelved Sora and walked back plans for a ChatGPT variant. A New Yorker profile of CEO Sam Altman generated the kind of buzz that PR teams dread.

Its work with the Trump administration and the decision to put ads in ChatGPT haven't helped the vibe either.

Bret Taylor, Sierra's co-founder and OpenAI's board chairman, defended Altman at HumanX when Bloomberg's Rachel Metz brought up the profile: "I think Sam's remarkable. I think he's a remarkable leader of AI, and I really trust his character." Taylor knows what the question really was. He answered it anyway.

OpenAI isn't collapsing; it's competing. That's different. The Wall Street Journal recently compared the two companies' finances and found both were growing faster than any tech businesses in history.

OpenAI this week launched a $100 ChatGPT subscription tier with deeper access to Codex, its coding tool. That move has Claude Code written all over it as a target.

So what's actually happening? The simplest read: OpenAI built a category and now has to defend it. Anthropic came in after, watched what worked, and built something a growing number of professionals seem to prefer for actual work. Not demos. Work.

Agentic AI is the thing everyone at HumanX kept circling back to: software that doesn't just answer questions but takes actions, runs tasks, and operates inside business processes.

OpenAI CTO of Business-to-Business [B2B] applications Srinivas Narayanan described the current moment as one where "every month, and sometimes every day, we are all looking forward to something new." That's genuinely true, and in that environment, reputation compounds fast.

Claude's appeal at HumanX wasn't about one feature or benchmark. It seemed more like a vibe shift; the feeling that a tool actually behaves predictably and doesn't do embarrassing things. That might sound like a low bar. In enterprise AI deployment, it isn't.

Whether this moment becomes a real lead or just a conference buzz cycle depends on what Anthropic does next; they've got the attention. Converting attention into lock-in is a different problem. OpenAI has resources to fight back; the new Codex tier is proof they're willing to.

But right now, if you spent time on that convention floor and asked people what they were actually using, the answer kept being the same one. That doesn't happen by accident.