Power up. From supercharging employee productivity and streamlining processes to inspiring innovation, Microsoft’s AI is designed to help you build the next big thing. No matter where you're starting, push what's possible and build your way with Azure's industry-leading AI. Check it out.
AI agents were the talk of the town as Silicon Valley leaders descended on the Big Apple for New York Tech Week. But some discussions still began with an exchange of definitions.
The decentralized conference consisted of more than 1,000 events, from pickleball matches and movie nights to panels and rooftop soirées, scattered throughout the city. This year’s iteration drew more than 60,000 RSVPs from attendees, according to one organizer.
A sizable chunk of those events focused on the rise of autonomous generative agents, billed as the next big wave in the AI arms race. Tech giants like Microsoft and Salesforce have gone all-in on this concept with lofty predictions about digital workforces at companies of the future.
But separating the hype from the current reality—or even nailing down a firm definition of what agents are—can be a tall order sometimes. Dan Balaceanu, chief product officer of enterprise conversational AI platform Druid, said buzz around the term has swallowed up some of the more precise terminology in the AI world.
“Somehow this agent name absorbed all the related terms,” Balaceanu told Tech Brew before he appeared on a panel on enterprise agents. “We don’t talk about virtual assistants anymore—we call them AI agents. We don’t talk about bots, we call them agents, we don’t talk about intents and skills or conversation flows, we talk about agents. In my opinion, maybe it’s not 100% accurate, but if we understand we are speaking about AI completing and helping humans, it’s fine.”
Coding agents: The slipperiness around what an agent is doesn’t seem to extend as much to software developers, who are already using agents every day, according to Zach Lloyd, founder and CEO of AI command line terminal startup Warp.
“The development stuff is in production and real and I’ve used it, whereas a lot of the other stuff, it’s more in demo-land or pitch-land, or even deck-land,” Lloyd told Tech Brew. “Every enterprise company has an agent product, but we’re not using any of them.”
Merrill Lutsky, co-founder and CEO of code review startup Graphite, said engineering might offer more clear-cut use cases for agents, whereas AI in other fields might be stymied by more varied processes or regulatory concerns.
“There’s a pretty clear drop-and-replace or drop-and-augment workflow there for AI to come in, versus…a combination of the models not being quite as powerful relative to humans yet in other domains, and that workflow not being as well-defined. And then also, obviously, you know, some regulatory concerns.”
Industry-dependent: George Mathew, managing director at venture capital firm Insight Partners, said on stage at an SAP panel on agents and copilots that AI has had the biggest impact thus far on three fields: coding, legal, and accounting.
“There’s a good, bad, and ugly to all this,” Mathew said. “The three most disruptive moments in AI right now—code, legal and audit, accounting—all three of those are being completely reimagined at the entry level.”
But Jared Coyle, chief AI officer for SAP, said on the same panel that even when enterprise technology is available to fields like sales, there’s a big challenge getting organizations to break from tried-and-true existing practices.
There are also plenty of technical hurdles when it comes to governing how agents interact. “As it turns out, there’s a lot of engineering aspects that you don’t think about,” Coyle said. “There’s orchestration considerations. There’s a lot of just validating who has access to what, and how do you define that from a governance perspective, and all of those have to be considered.”
When AI does make a big business impact, however, you may not see it in headline news, according to Coyle.
“The best AI is boring,” Coyle said on stage. “We’ve done a quantitative evaluation of the 34,000 or so organizations that are actually using our AI. And what’s interesting is you can make [that] one conclusion from it…It’s not the ones that you see on the keynote stage. It’s the ones that you sit back and you go, ‘Oh, that’s actually the most impact that we’re seeing and the most uptake.’”