Why Edward Jones’ agentic AI trial comes with limits

These internal AI agents function as “always-on digital teammates” for the company’s branches, she said. Their capabilities include things like monitoring workflows, pulling and automatically summarizing information, drafting content, generating recommendations and initiating next steps.

However, Widhani isn’t ready to hand over the reins to AI agents yet. Instead, human employees are asked to “refine and turn in finished work,” she said. Widhani will review these agentic AI partnerships by the end of the year to determine whether to commit long-term, given AI is moving faster than marketers can keep up with, she said.

“We take this approach because current AI tools cannot yet deliver the nuance, emotional clarity, and sense of humanity our brand requires,” Widhani said.

‘AI is an adoption journey’

As the pilots are ongoing, Edward Jones has not yet set goals for the agentic AI itself outside of the general efficiency play. There are goals, however, for the majority of the marketing organization to be AI enabled by the end of the year, per the exec.

“I’m not certain whether that happens or not, but you’ve got to set really high ambitions on these things because I do think AI is an adoption journey,” she said. 

Rocket Companies, parent to companies such as Redfin and Rocket Mortgage, has similar ambitions. All creative team members have a general mandate to use AI 20% of the time, said Jonathan Mildenhall, CMO at Rocket Companies. When it comes to agentic AI, however, the intent is to continue to test and learn for all the same reasons often stated by other CMOs at present.

AI ambitions vs. reality

Agencies are in the same predicament, still in the testing and learning phase for agentic AI. It begs the question as to whether these tools are truly agentic or “buzzword de jour” for any interaction with AI, said Mark Wahl, vp of technology and innovation at White64, an independent agency.

It’s not that the tools aren’t available. For example, NBCUniversal started testing AI agents to sell ads against a live NFL game in January. It’s a matter of trust until marketers are able to bridge the disconnect between AI ambitions and reality. 

“It’s going to have to prove itself that it’s going to act consistently, reliably and safely before we’re really comfortable letting it run on its own,” said Wahl.

Agencies don’t see agentic AI’s dependance on human intervention as a setback. Notably, the advent of AI has resparked conversations about in-housing and questions as to whether AI will undermine client-agency relationships. Agency execs position AI as a tool that achieves faster results alongside clients.

“For us, agentic AI is about training goal directed systems that help us achieve better work, at a faster pace,” Phil Camarota, chief creative officer at Flywheel, said in an emailed statement to Digiday. He added, “I do not view the human oversight as a limitation.”

While the question remains as to whether agentic AI is truly capable of functioning independently, it hasn’t dampened marketers’ AI appetites. The numbers show it. Agentic AI is expected to resolve 80% of common customer service issues, sans human oversight, by 2029, according to Gartner research. Seemingly, marketers are still squaring what agentic AI looks like embedded in more tedious processes — like media buying, campaign optimization and content creation.

“My ambition, though, is for a large part of the marketing team to be leveraging AI. I have put out there that I want a majority of our organization to be AI enabled by the end of the year,” said Edward Jones’ Widhani.