娇色导航

In ServiceNow’s second annual , over 4,500 worldwide private and public sector leaders were surveyed and findings revealed that this year’s average maturity score actually dropped from last year, from 44 to 35 (out of 100 points). In addition, fewer than 1% of respondents scored over 50 on their 100-point AI maturity scale. But this is, in fact, good news for the industry and for AI adoption and here we explore why that is and how CIOs are moving forward.  

In a previous article, 4 recs for CIOs as they implement agentic AI, I noted that despite the hype, CIOs agree there’s an approaching reset of agentic AI expectations. The ServiceNow report is further confirmation that the reset is underway, which is due to a number of factors. In implementing agentic AI, organizations began to know what they didn’t know. Unlike gen AI, which could be implemented as a standalone or bolted on somewhere, agentic AI requires far deeper integration — at least if you want to utilize it for maximum benefit.

Instead of just throwing up a user interface and asking questions of a gen AI chatbot, CIOs are looking to use agentic AI to execute tasks and orchestrate workflows going deep into enterprise processes, such as CRM, supply chain, enterprise resource planning, HR, finance, and more. All this, plus the pace of tech advancement, existing silos and legacy apps, and hundreds of new agentic AI use cases, means seriously upping their game in terms of AI integration, orchestration and governance, keeping humans in the loop where needed and taking it beyond pilots to production-grade implementations.

I spoke with several tech leaders to find out what they’re seeing in the market and how organizations are dealing with their agentic AI implementations. Here are four approaches they’re taking into the enterprise.

1. Dealing with and managing exceptions

As agentic AI gets implemented within core enterprise processes, it needs the ability to deal with and manage business exceptions — something that prior software and automation techniques such as RPA have often struggled with.

“It’s still early days, but the impact of AI agents is tangible and growing quickly,” says Dan Priest, CAIO at PwC US. “AI agents are showing real promise, especially in areas like supply chain and financial operations, which have historically required significant manual effort to handle exceptions. AI agents are quite good at dealing with complexity, including exception handling, due in part to their ability to bring the insights of how similar exceptions were resolved into real time scenarios.”

PwC’s recent found that 83% of executives said agents were helping solve business problems faster and more effectively. For example, in supply chain, they’re seeing AI agents monitor real-time inventory levels, flag anomalies, and autonomously trigger reorders, cutting down on delays and reducing human bottlenecks.

2. Pursuing a multi-agent approach

EDB’s recent on how the world’s largest enterprises are preparing for the agentic economy, and their ongoing conversations, shows that the most successful of them deploy 10 or more agentic agents and business specific gen AI intelligent applications.

According to Nancy Hensley, CPO at EDB and EDB Postgres AI, these organizations have spread agentic outside traditional customer support to supply chain, marketing content, automation, and even digital twins. “They do twice as much as all the other groups because they recognize there’s an ROI cap in each business application, so they do as much as possible,” she says.

When implementing agents, Priest is also seeing a strong commitment to responsible AI. Clients want to see AI benefits, and they largely recognize the importance of achieving those benefits responsibly. “They’re leveraging AI agents to empower their people,” he says. “This means putting proper guardrails in place, preparing teams to work alongside agents, and building high-performing groups where humans and AI collaborate. Ultimately, it’s about fostering a human-AI partnership where each does what it does best, and both evolve together over time.”

3. Rethinking architecture

CIOs are also rethinking their enterprise architecture and combining the best of both worlds in terms of using AI agents from their existing enterprise application vendors, as well as introducing agents from outside pure-play vendors. “Many clients are starting with AI agents offered by their existing ERP or CRM vendors because they’re natively integrated and relatively easy to implement,” says Priest. “At the same time, there’s growing interest in purpose-built, task-specific agents from outside vendors that can handle more complex or specialized workflows. And as multi-agent solutions grow in popularity, orchestration platforms that help harmonize across agent platforms play an important role in agentic workflows.”   

The concepts of sovereign data and AI platforms, as well as AI, data, and workflow platforms, are becoming increasingly popular, too. “The number one predictor of success for top performers in our research was the view that being their own sovereign data and AI platform was the foundation for that success,” says Hensley. “It allowed them to be secure and expansive in delivering this future, now.”

4. Refocusing on innovation

Moving beyond efficiency is vital for top performers when working with agentic AI. The most successful organizations are ambitious, but not just efficiency focused. They’ve seen the technology prove out, and they want to scale it. They’re thinking strategically about what the benefits are, where to drive change, and they’re setting measurable goals that can go all the way up to the board.

“Taking time, cost, and touches out of work is valuable, but leaders are refocusing on innovation,” says Priest. “Where can they combine their differentiators with AI to create new sources of advantage in the market? ROI from new AI powered efficiencies is quickly becoming the new standard for table stakes.”  

How the reset will advance your maturity

Of course, these approaches aren’t the only steps forward-thinking CIOs and CAIOs are taking when implementing agentic AI deep within the enterprise. But they provide hints of where to focus. Think about dealing with and managing exceptions, core workflows, and an orchestrated multi-agent approach with responsible AI guardrails and human-in-the-loop processes wherever needed. Also think about new architectures that combine AI, data, and workflow in a sovereign platform. Then combine your differentiators with AI to create new sources of advantage in the market.

Your agentic AI reset in terms of time spent on these approaches and others will yield reward and propel your organization forward in AI maturity.