CxOs in the BT150 are starting to push back on agentic AI washing from enterprise vendors.

Our April BT150 meetup touched on multiple topics including tariff volatility on projects and costs (everyone flying blind) and ascending cloud vendors (Google Cloud and Oracle Cloud Infrastructure have momentum), but agentic AI again was the big theme.

Agentic AI has been a running topic among the BT150 and the biggest issue is discerning what's real, what's hype and are the data ducks in order at their enterprises. Our CxO call, which is operated under Chatham House rules, summed up the skepticism about the short- and mid-term returns on agentic AI. Overall, CxOs buy into AI agents, but you can quibble all day about timelines.

2025 BT150 zeitgeist:

One CxO summed up the state of agentic AI: "I have struggled to find an agent that wasn't an API that was just wearing an agent t-shirt. I'm still struggling to find one." 

That take is worth a mention because last BT150 call there was a few questions about whether RPA is a way more cost-effective approach to some of these issues. CxOs said that vendors are often touting agents when it's click automation or just an interface that doesn't scale.

There also needs to be a lot more work on API automation, finding the right ones and connecting APIs while negotiating credentials and data exchange.

The big takeaway is that CxOs need to beware of agents that are just dressed-up APIs.

"We had a lot of cloud washing. We had a lot of AI washing. And now we have a lot of agent washing," one CxO said.

Constellation ShortList™ Cross-Platform Agentic AI

Other tidbits on agentic AI include:

  • Horizontal players are likely to be your agentic AI most valuable players. Integrators and consultants--Capgemini, Cognizant, Infosys, Accenture and the like are going to be better at building agents than vendors who aren't going to resist the urge to keep you on their platforms. ServiceNow and hyperscale cloud providers would also fit into those horizontal plays.
  • Boomi got a few shout outs as a layer to coordinate APIs and agents. In some ways, it's a twofer in that Boomi can manage APIs that are dressed up as agents if that's the shorter-term play.
  • Shadow AI agents are going to be an issue. If every business user and department can create an agent you're going to have governance headaches real soon.
  • In the meantime, make sure you have your data lakehouse architecture in place as well as your data foundation.
  • CXOs are finding niche consultants focused on agentic AI that are following a concept called "unconsulting" where it's a 90-day sprint or less to build agents on a few already identified use cases. Think of T-Mobile's uncarrier approach but for agentic AI.
  • Agents are a way of getting out of generative AI jail where there was a proof of concept that didn't scale and got held up in compliance.
  • Focus on the processes and the optimization so you can automate with AI agents.