What enterprise leaders are most excited about with AI right now

May 31, 2026

Notes from a dinner with the people actually deploying it.

This week I hosted the Enterprise AI and Agentic Security Dinner in Los Altos Hills, a room of the people who decide what AI ships inside their companies and who own the risk when it does. CFOs, CTOs, security leaders, founders. The event runs under Open Future Forum, where I spend most of my time getting operators in the same room off the record. I moderated a panel with Manjush Menon, Karthik Suresh of ZoomInfo, and Paulina Xu of Agentic Fabriq, covering identity and permissions for agents, governance and standards at scale, and what leaders should prepare for over the next 12 to 24 months.

I went in expecting caution. There was plenty. What stayed with me was how excited these people have become, and how specific that excitement now is. A year ago the conversation was about potential. This week it was about results.

Here is what they are most fired up about.

Agents are doing real work, not demos

The party trick phase is over. The leaders in the room were not talking about what AI might do someday. They were talking about workflows running in production right now. Claims routing. Customer onboarding. First-pass code review. Support triage that clears the queue before a human ever sees it.

The shift in tone was the tell. People stopped asking whether agents work and started comparing notes on which processes pay off first. When a CFO can point at a line item and say a process that took three people now takes one and a review, the argument is over.

Deployment, the hard part, is finally getting easier

For two years the blocker was not the model. It was everything around it. How do you give an agent real access without handing over the keys. How does a regulated company say yes when the legal team is in the room.

That gap is closing fast, and the people closest to it are the most optimistic about why. On the panel, Manjush Menon made the case that governance and structure are not afterthoughts. They are the foundation. Clear use cases, defined workflows, scoped permissions, humans in the loop where it matters, and an honest read on what the model actually produced. That, he argued, is what makes AI adoption hold up at scale.

It reads as optimism once you see the tooling catching up to it. Paulina Xu’s company, Agentic Fabriq, is built around giving agents identity and scoped permissions, so an agent gets access the way an employee does, not a blank check. Control planes now let a company govern every agent from one place. Execution can run inside a company’s own perimeter instead of someone else’s cloud. None of this makes headlines. All of it is the reason a bank or an insurer can now move from pilot to production.

Small teams are doing the work of large ones

The optimism that surprised me most was about people, not technology. More than one founder talked about shipping at a pace that would have needed three times the headcount two years ago. Engineers are spending less time on the work nobody wanted and more on the work only they can do.

Nobody in the room framed this as cutting people. They framed it as range. A small team can now take on problems that used to be off limits. For an early company, that is the whole game.

They can see the edge now

The clearest signal of all was confidence about direction. A year ago the honest answer to “where does this go” was a shrug. This week the leaders deploying AI could describe the path. They know which problems to point agents at first, what good deployment looks like, and where the moat forms.

That clarity is its own advantage. The companies that figure out where agents help are already pulling away from the ones still debating whether to start.

The takeaway

The mood in enterprise AI has changed. The caution is real and the security questions are serious, but underneath them is something I have not seen at this level before: leaders who have stopped wondering and started building. They are excited because the work is paying off, the tooling has caught up, and the path forward is finally legible.

The best conversations on this still happen in person, off the record, with the people doing the work. That is the room I try to build. If you are an operator deep in this, come find me.

Murray Newlands is a Partner at IA Seed Ventures, Partner at Tilden Family Office Group, and founder of Open Future Forum. Subscribe at substack.com/@murraynewlands.