All eyes on AI agents for business in 2025, leaders say

DAVOS, Switzerland – Businesses need to strategize their AI plans, particularly in developing a process for AI agents, Writer CEO and co-founder May Habib told Axios' Alison Snyder.

Why it matters: Some companies may be lacking proper guidelines and missing competitive opportunities, as tech giants are pointing to AI agents as the next big thing in 2025.

Axios' Ina Fried and Snyder moderated conversations with Deep Learning AI founder Andrew Ng and Habib in Davos. The Jan. 21 discussions were sponsored by Cisco.

What they're saying: Habib said businesses don't want AI agents making too many decisions, but need to figure out the fine line between what needs human approval and what doesn't.

• "What we see as really the bottleneck here is getting people and companies and teams to really align on 'what is our process,'" Habib said.
• "The tech building blocks to execute on that exist today, but the big body of work" is the gap in skills, she added.

More businesses are now seeing positive ROIs from their AI investments, Ng noted.

• "If you're not in the business of training foundation models, because others have spent billions of dollars training these models, you can now get access to these models for cents or dollars … and build really valuable applications on top," Ng said when discussing applications like AI agents.
• "So the ROI equation for people building applications actually looks really good, even though it still needs to be sorted for the people building foundation models."

2025 prediction: Habib said she expects enterprises using AI will have "built their own first [application] or agent" by this year and added that the acceleration is happening "pretty fast."

Sponsored content:

In a View From the Top sponsored segment, Cisco EVP and chief customer experience officer Liz Centoni said AI agents aren't all hype, but it will take a "long time before it becomes a reality" because AI agents exacerbate some current AI issues.

• "When we think about the issues that we've been talking about with just AI in general around transparency…those just get even more exacerbated with agentic because with agentic, … it's like, you're talking about your intern [who] is now actually graduating to a full-time job," Centoni said.