Enterprises chasing agentic AI are likely to run into a reality check, but there’s considerable upside ahead if they can navigate the moving parts.
At Constellation Research’s Connected Enterprise 2025, CxOs discussed the current state of play for AI agents, the reality and what’s ahead.
The week in CCE 2025:
Here’s a look at the AI agent game from CxOs.
The dream
Pick through the hype of agentic AI and the various vendor pitches and the dream revolves around autonomous enterprises that simply flow. They’re living, breathing businesses that aren’t siloed and can evolve, react and get things done unsupervised.
Salesforce evangelist Vala Afshar is working on a book about becoming autonomous. At Constellation Research Connected Enterprise, Afshar said:
“Autonomous machines will have not just situational awareness, but horizontal awareness, marketing, sales, commerce, every department, once you become an agent, enterprise, will eventually create that anticipatory muscle you have where one signal can really activate your business to have beyond situational awareness. When you become autonomous, you resemble living organisms. All living organisms are flow based.”
Business flow comes from culture, roadmaps, customer engagement and ecosystems. “Be mindful of flows in and out of your business,” said Afshar.
The working theory is that AI agents and the orchestration platforms will keep these flows going. Ultimately, enterprises will use agentic AI as a circulatory system for processes and business operations.
The reality
If there was one takeaway from Connected Enterprise 2025 it’s that the reality behind AI agents is vastly different than the dream.
Simply put, we’re not there yet but may be in a year. CxOs don’t have a comprehensive strategy, are lining up data to be AI friendly, overwhelmed with platforms and in many respects would prefer to build systems that recapture the funds that go to SaaS vendors.
Platform fatigue is setting in.
Fiona Tan, CTO of Wayfair, said the platform pitch from SaaS is something enterprise customers need to be wary about. She said:
“We’re looking for a horizontal platform partner that we can work with. Some of those integrations we will do directly, and then some of them we will look for with SaaS partners. The difficulty right now with SaaS is that they’re also trying to go horizontal. We want to access enhanced capabilities of each of the SaaS partners, but control it. We may not want that necessarily.”
The AI agent strategy conversations are happening. Are they fully baked? No. They’re getting there.
AI agents need a manager
Boomi CEO Steve Lucas said that orchestration is going to be more important than creating AI agents. Like human workers, oversight will be critical.
Dr. Janice Presser, Founder of Teamingscience.com, said deploying AI agents needs the same thought behind it as hiring a human.
“Whether you’re hiring a human being or a carbon based life form or silicon based life form, you need to kind of narrow down how you want them to be and what you want,” said Presser.
Presser noted the following:
- AI agents will team with humans.
- Both types of workers will need to be managed differently. “Agents are for work. Humans are for love,” said Presser.
- You’ll need to pick the right tasks for agents just like you would for human employees.
- Since you’re building a team you will need AI agents that complement each other and act in different ways. You can empower AI agents to act like a human with creativity or simply handle tasks.
“At the end of the day, you have to have these agents talking to each other and coordinating. There’s no getting away from that,” said Babak Hodjat, CTO of AI at Cognizant.
The journey is exhausting
The rate of change for AI is exhausting and there’s so much flux CxOs are wary about making any big bets on platforms as well as models. Flexibility is the key word.
Not surprisingly, build vs. buy was a key debate. The consensus among the Constellation Research community was more in the build camp. Ultimately, enterprises are going to do both build vs. buy.
In other words, there’s no easy button for AI agents and unless you have your data house in order it’s hard to chase the dream.
An industry like banking, financial services and insurance is regulated can scale into AI agents because it has historically leveraged data for competitive advantage.
“This is a perfect opportunity for a huge amount of automation, which we can then augment with a human in the loop,” said Sudhir Jha, Founder & CEO of Golconda AI. “Instead of trying to solve one particular problem, think about the entire flow and how to make the entire process more efficient using AI agents.”
What’s next?
AI agents are going to create multiple challenges beyond the technology issues. Here’s a look.
- Leadership challenges. Patrick Naef, Managing Partner Boyden Global Executive Search, said there will be leadership challenge due to AI. “We’ve trained over the last decade to move away from hierarchical cost designs to empower people, sharing missions, strategies, cultural goals, and then the people decide what to do,” said Naef. “Will that work with AI or do we need to redefine hierarchical structures and still empower people. That’s quite a challenge. The interesting thing is how we’re going to talk about passion when centralizing that one process and workflow.”
- Develop a high-level strategy. CxOs repeatedly argued that the AI agent movement can result in a sprawl of vendors, platforms and digital workers.
- Think organization. Corregan Brown, Director of Engineering, CTS Restaurant Experience at Chick-fil-A, said AI agents and the future of work will revamp organizational structures. “Organization is going to go from caterpillar to butterfly. There will be some major org that was structured in some way that made sense to us, and it’s going to come out incomprehensibly different next year,” said Brown.
- Work on the building blocks that can enable AI agents. There will need to be platforms, orchestration tools, AI agent builders and data strategy. Enterprises are in various stages with those efforts.
- Realize that you may be overthinking the AI agent play. On an HCM panel, Constellation Research analyst Holger Mueller said AI agents won’t be on the org chart and shouldn’t be. He said: “My view is that agents are just machines. Agents are run by hardware and software. They’re just machines.”
More:
