Editorial
As the AI landscape has evolved, the conversation around GenAI has taken an interesting turn toward “AI Agents.” In fact, Gartner called this the top strategic technology trend for 2025. These autonomous agents represent a new frontier in AI, equipped to perform complex tasks, make decisions and even learn over time.
But with this promise comes a mix of opportunity and challenges.
In this article, I will delve into the emergent capabilities of AI agents, examine their most compelling use cases and explore the potential risks, including security concerns raised by leading industry voices like Okta’s CEO. We’ll also consider the critical role of CIOs in facilitating the adoption of AI agents and the key partnerships they must foster to successfully integrate this technology.
What CIOs Really Think About AI
Let me start this article by saying that CIOs having strong reactions to AI agents reflect a balance of skepticism, optimism and realism about their current and future potential:
- Jonathan Feldman, CIO for Wake County, is worried AI agents are being overhyped, noting that the exciting results from generative AI deployments haven't materialized yet.
- Joe Sabado, UC Santa Barbara deputy CIO, expressed personal skepticism about AI agents, highlighting that while exploring possibilities is enjoyable, he hasn't yet found highly effective uses for AI agents in his own experience, aside from tools like the Replit Agent.
- Peter Salvitti, chief technologist for Boston College, noted from a Gartner SYM session that only 48% of AI initiatives have met or exceeded expectations, indicating a gap between anticipated and actual business outcomes.
- Tim Crawford, former divisional CIO for Konica Minolta, is even more critical, stating that most AI initiatives have a lower success rate than reported, with many steps in the AI process needing to work seamlessly for projects to succeed. He emphasized that while the future of AI holds significant promise, success will require hard work and attention to risks like privacy violations and disinformation.
- Isaac Sacolick, former CIO at McGraw-Hill Construction, now under Dodge Construction Network, finds value in large language models (LLMs) but stresses that they require expertise to maximize their effectiveness. He sees great promise in AI's ability to transform areas like HR and customer support but noted that AI sprawl, particularly with agents, will present governance challenges for organizations.
Related Article: Has AI Delivered on Its Promises?
Exploring the Potential of AI Agents in the Digital Age
Generative AI has without question evolved rapidly, and now a key focus is upon AI Agents. CIOs are clear that AI agents are no longer on the periphery — they’re integrated into every major SaaS platform today. For example:
- HR agents are streamlining recruiting processes
- CRM agents are personalizing outreach and messaging
- Contact center agents are enhancing customer interactions
Given the potential, CIOs are seeing employees experiment with these tools in some capacity, said Sacolick. At the same time, the market is expanding with developer tools aimed at building AI agents.
Some platforms offer purpose-built solutions to import data directly into agents, while others allow for customization based on existing data sources. However, Sacolick argued, “despite the proliferation, none of these solutions are yet transformative.” Most applications of AI agents enhance tasks that employees are already performing while others assist with occasional tasks, like submitting expense reports.
Nevertheless, AI agents, according to DUNELM Associates CIO and managing partner Martin Davis, “are creeping into almost everything, whether you want it or not! And yes, your people are probably using them in all sorts of tools.” At the same time, it’s highly likely that employees are using them in ways that are flying under the radar. Engaging with vendors and discussing practical AI agent use cases has become the catalyst for conversations for how these tools can be harnessed effectively.
One area generating considerable attention is reengineering the Software Development Life Cycle (SDLC) for agent-based approaches, particularly in fault/root cause identification. While these applications are promising, stability remains a significant concern.
CIOs pointed out that the term "agent" has been used so broadly that it can effectively describe everything from a simple HR chatbot to highly complex, mission-critical systems. Understanding and navigating this expansive landscape is crucial as AI agents continue to evolve, potentially disrupting traditional workflows and roles.
Compelling Use Cases for Agentic AI
I asked CIOs what use cases are most interesting? What is most transformative for the business? Sacolick said, “agents can dramatically change how people work and drive productivity while expanding what employees can do.”
Use cases that stand out include personal shopping agents for ecommerce, which could enhance customer service, and AI agents in manufacturing and healthcare could streamline operations and support critical decision-making.
By presenting relevant information, connecting to APIs and automating tasks, agents can reduce workloads, boosting productivity and enabling employees to focus on higher-value tasks. However, the scalability and integration of these agents across enterprise functions introduces new risks, particularly when something goes wrong, making them potentially more impactful — but also more vulnerable — than traditional apps.
Meanwhile, Davis is clear, “it is still early in the hype cycle; the usefulness and degree of help varies dramatically. There [are] a lot of time saving opportunities. The more interesting ones will be ones that make a quantum difference, not small incremental ones.” In contrast, FIRST CIO Deb Gildersleeve said, “I haven't seen many ideas that are transformative yet, but it is still worthwhile to identify smaller use cases to get started.”
As the technology develops, early adopters are focusing on small, incremental improvements, but the true potential lies in more transformative use cases. Agents that combine machine learning with specialized domains, such as causal AI or AI integrated with physics models, could drive quantum leaps in efficiency and autonomy. These innovations could be particularly transformative once trusted enough to operate with minimal human intervention. While the market is still early in the hype cycle, larger breakthroughs are on the horizon, and they hold the most promise for fundamentally altering how businesses operate.
Much of this agrees with the perspective of Jonah Midanik, general partner & COO of Forum Ventures, who said, “In the near term, I believe almost every use case will be impacted by AI, but sales are likely to see the most significant change first.
“Today, sales teams spend a large portion of their time on research, outreach and follow-up, rather than face-to-face interactions with customers. It’s easy to envision AI handling much of that work, allowing salespeople to focus almost entirely on direct human engagement and solving customer challenges. The hiring process is another example, where repetitive tasks can be efficiently managed by AI, leading to significant shifts in the people and sales functions. These areas will likely be the first to experience dramatic change, with the impact resonating across the broader enterprise as adoption increases.”
Related Article: Navigating the AI Landscape: A CIO's Guide to Success
Realistic Risks That Come With AI Agents
Okta’s CEO Todd McKinnon has recently, in a CNBC interview, raised a crucial point about the potential of AI agents falling into the wrong hands.
The risk of hackers taking control of these agents is certainly real. While hacking has always been a threat, the ability for malicious actors to inject instructions into autonomous agents adds a new layer of vulnerability. These agents, connected deeply into business functions and enterprise systems, can potentially be exploited in ways that amplify the consequences of a cyberattack. A hacked agent might not only carry out a single malicious task but could continue operating undetected, leading to widespread damage.
However, while the risk is real, its likelihood and impact depend on several factors, including how well companies manage and secure these agents.
As Midanik suggested, “there are quite valid concerns about potential security risks, but it’s important to recognize that hacking is already a constant threat. The worry about hackers injecting instructions into autonomous agents is legitimate, but the risk can be managed by maintaining human oversight at critical points. While new technologies bring new risks, these can often be mitigated, just as we’ve done with past innovations. The evolution of technology will always introduce fresh challenges, but these are both inevitable and addressable.”
It’s also worth noting that current AI agents, like LLMs, are not immune to issues like hallucination or delivering incorrect information. The network effects of these errors can be catastrophic, particularly in industries where accuracy and reliability are paramount. Until these risks are fully addressed, there is valid concern that agents aren’t yet ready for use in high-stakes environments where subversion could be disastrous. As these tools become more embedded in key business processes, their exposure to sensitive data will only increase, making security even more critical.
Webinar
Oct
22
Beyond Storage: Smarter Content, Bigger Impact with DAM + AI
Discover how the DAM + AI duo makes content smarter, stronger and more accessible.
Register
Webinar
Nov
6
Fix the Content Bottleneck: Build a Better WebOps Strategy
Content stalled? Dev overloaded? You’re not the only one. Learn how streamlined WebOps bridges the publishing gap.
Register
Webinar
On demand
Agentic AI Playbook: Real-World Customer Service Use Cases You Can Deploy Now
Boost self-service by 30% and slash call volume by 63% with agentic AI.
Watch Now
Webinar
On demand
CMS Briefing: A Live Look at What’s Next in AI-Driven Platforms
Learn how leading organizations are using AI‑driven tools to publish faster, personalize smarter and stay secure.
Watch Now
Webinar
On demand
Ready or Not: How Data-First Organizations Are Unlocking Agentforce Potential
Learn how to cut through the noise, activate Agentforce and build a Salesforce AI strategy that actually delivers.
Watch Now
Webinar
On demand
AI in Customer Service: Faster Resolutions, Happier Customers
Don’t let rising demand burn out your team. See how to build a smarter, more resilient support org.
Watch Now
Steve Jones, Capgemini executive vice president, data driven business & GenAI, said, “100% the current generations are massively open to subversion or complete control. Like LLMs ‘hallucinate’ this is a network effect of hallucinations. Not ready for use where subversion is a concern. Boundaries matter, you can’t build towards one great big blob that knows everything, far too much risk.”
Worse yet, Joanne Friedman, CEO and principal of Connektedminds, argued, “a hacked agent will be tougher for a user to identify.”
The CIO’s Role in AI Agent Adoption & Deployment
CIOs have a pivotal role in shaping the adoption and safe deployment of AI agents within organizations. One of their key responsibilities is to ensure there is a standardized approach to modeling problems so that AI agents can be implemented securely. This includes promoting proactive data governance, particularly for unstructured data, and addressing existing data debt that may hinder the effective use of these technologies.
As AI agents increasingly integrate with critical business processes, CIOs must take the lead in red-teaming efforts to identify vulnerabilities, ensuring robust cybersecurity frameworks are in place to mitigate risks.
Partnerships are essential in this context, especially with data management teams and cybersecurity experts. CIOs should also build strong relationships with vendors providing AI tools to ensure a clear understanding of data lineage and provenance. Trust in the data that fuels AI agents is paramount, and CIOs must maintain transparency and control over what agents are experimented with and deployed. Collaborating with legal and compliance teams is also critical to navigate regulatory challenges and ensure ethical use of AI agents in the organization.
Related Article: Why CIOs and CDOs Need to Work Together on Generative AI
CIOs: Prepare Now for Future AI Advancements
AI agents represent a transformative opportunity, but they also come with significant challenges, particularly around security and governance. CIOs must play a central role in navigating this complex landscape, driving innovation while safeguarding their organizations from emerging risks. Key partnerships with cybersecurity teams, data management experts and AI vendors will be crucial in ensuring that AI agents are deployed responsibly.
As AI agents continue to evolve, their potential to automate tasks and enhance productivity is undeniable, but success will depend on how well CIOs can balance innovation with risk management.
Learn how you can join our contributor community.
