Preparing your workforce for AI agents: A change management guide

Many CIOs struggle to develop change management programs in their digital transformation initiatives. Leaving change management as an afterthought is a costly mistake for CIOs who need employee adoption as a key step to delivering business outcomes.

Rolling out AI agents doubles the importance of change leadership and increases the risks when transformation initiatives fail to shift behaviors. Early AI adopters may spur the use of rogue AI agents, while employees who fear that AI agents may take their jobs can become detractors.

Recent research reports highlight gaps in AI change leadership programs. According to recent report from MIT, 95% of organizations are getting zero return from their AI investments. One possible reason is that only 14% of organizations have a change management strategy, according to an AWS report.

“The best technology delivers zero value if no one uses it, and adoption is the final, critical mile,” says Michael Connell, COO of Enthought. “Leaders must not only budget for change management as seriously as they budget for building, but also involve end users who are going to be the consumers of the technology early and consistently in an agile development process.”

Lead change management at AI’s speed of innovation

CIOs should first recognize that rolling out AI agents is nonlinear, requiring several parallel change management efforts. The rapid evolution of gen AI technologies is one reason for this, making what was hard or expensive to implement just a few months ago more feasible and less expensive today.

What’s not getting easier, however, is achieving alignment around AI strategy, implementing robust governance, deciding which areas of the business to experiment in, and transitioning AI capabilities from pilots to production.

Aligning on terminology is one important place to start. In my review of AI agents from leading SaaS and security companies, the term is broadly defined to include natural language and reasoning AI models embedded in workflows and in the tools that support employee and customer experiences. Many AI agents support a single job function, but capabilities vary significantly. Most AI agents aren’t fully autonomous or fully plugged into agent-to-agent workflow orchestrations, but their sophistication is likely to advance.

Define workforce segments based on AI responsibilities

A CIO’s first objective is to break down the organization into segments related to AI responsibilities from investment through adoption. These segments should include:

• Executives involved in aligning on strategy, defining business outcomes, and setting investment priorities
• Compliance leaders in risk management, information security, and data governance responsible for evolving the AI governance framework
• Subject matter experts across all operating functions, who provide the required knowledge for AI agents and will be most involved in validating their accuracy
• End users who will be using AI agents as a way to get work done, especially in areas that deliver greater strategic value beyond productivity improvements
• Innovators forming agile, cross-disciplinary teams who will lead experimentation, evaluate AI agents from different technology providers, and lead the development of proprietary AI agents

Using these segments, leaders can align change management programs more closely with how employees will participate in AI agent initiatives.

Guide executives toward strategic priorities

With all the hype around AI, many CIOs face a fleet of general managers and department heads vying for pole position in the AI strategy and investment priorities. But for many CIOs, AI isn’t their first transformation journey, and they know that spreading resources to support everyone’s wish list is a recipe for low-impact results.

Brandon Sammut, chief people officer at Zapier, recommends: “Anchor your AI agents imperative in two to three opportunities to boost existing priorities and goals. That keeps AI agents at the center of the company’s focus, and avoids the ‘sideshow’ trap that plagues most technology transformations. Start with a clear ‘why’ and ‘why now’ to shore up cultural foundations because embracing a new technology is a pressure test of an organization’s health.”

Recommendation: CIOs should recognize that they can’t be the only change leaders driving executives to align on a few strategic priorities. CIOs leading world-class IT departments consider executive relationship development a core competency. They identify ambassadors to meet regularly with department leaders, develop listening skills to identify force-multiplying opportunities, and draft vision statements for the most promising ones. These practices help evolve the AI strategy, and then change leaders turn up their communication skills to gain executive alignment.

Collaborate on AI governance before experimentation

In virtually every wave of disruptive technologies, the race to innovate, deploy, and gain strategic advantages well outpaces the organization’s ability to mitigate risks, define policies, and establish security guardrails. CIOs have the daunting task of bringing compliance leaders and AI governance to the forefront without stifling innovation and experimentation.

“Organizations rushing to deploy AI agents often overlook the gap between prototype success and production-ready systems, where traditional safety controls fail against reasoning agents that can bypass standard safeguards,” says Kamal Anand, president and COO of Trustwise. “The missing elements include embedded trust frameworks, real-time governance tools, energy-efficient infrastructure, and skilled professionals who understand dynamic agent behavior rather than static AI models.”

Governance teams will also need to develop risk frameworks for AI agents’ decision-making authority, as well as metrics for assessing the quality of their recommendations.

Elad Schulman, CEO and co-founder of Lasso Security, says, “CIOs must define which tasks AI agents can perform independently and which demand human oversight, especially when handling sensitive data or critical operations. No AI agent should run in production with full privileges before completing thorough security assessments, adversarial testing, and sandboxing.”

Recommendation: The change requires CIOs to act as facilitators, first with compliance leaders to prioritize guardrails aligned with material risks. Then, CIOs must oversee a communication strategy to ensure the entire organization understands which AI tools are permitted, the appropriate uses of enterprise data sets, and other AI compliance policies.

Champion subject matter experts who embrace AI

Boobesh Ramadurai, vice president of gen AI capability development and marketing analytics at LatentView, says, “If your processes rely on tribal knowledge, scattered data, or manual decisions, agents will stall.”

It’s already challenging for new employees to contribute when knowledge is hoarded or business processes are riddled with undocumented exceptions. It will be doubly challenging to introduce AI agents if subject matter experts are detractors and slow to share knowledge or provide feedback on an agent’s performance.

“Codify your business logic so agents can follow it by standardizing metadata, defining escalation rules, and ensuring systems are connected,” Ramadurai recommends. “Upskill teams to focus less on execution and more on orchestration, while analysts should design live feedback loops.”

AI agents have language and reasoning models, but their responses and recommendations are non-deterministic. Testing them will require subject matter experts to review and validate agent performance, provide explanations on errors, and suggest improvements.

 “AI agents won’t always do exactly what you tell them to, especially as they begin to make decisions based on patterns in data rather than explicit instructions,” says Dave Killeen, field vice president of product at Pendo. “Teams need to understand how agents are interacting with real workflows, catch when outputs drift from expectations, and have clear ways to intervene when needed.”

Recommendation: Subject matter experts play an important role in ensuring that AI agents act responsibly and are trustworthy. CIOs should partner with HR to define performance objectives and incentives for experts who embrace AI and contribute to successful programs.

Reward end-users with learning opportunities          

Cindi Howson, chief data and AI strategy officer at ThoughtSpot, says, “Workers are fearful of AI replacing them right now, so job one for leaders is to address their fears and map a plan for reskilling, which includes AI literacy, where today, 88% of Americans fail.”

Employees’ AI concerns should be at the forefront of CIOs’ change management programs. Communications and programs should be tailored to different job functions and skill levels, especially in areas where the adoption of AI agents by employees is prioritized.

“AI is definitely changing the workforce, and yes, we’re seeing the biggest impact on entry-level jobs,” says Geoffrey Godet, CEO of Quadient. “What often gets missed is that AI replaces tasks first, not people, and that opens the door to redesign roles in smarter ways. The companies that will thrive are the ones investing in upskilling and human-AI collaboration, because that’s where you unlock creativity, culture, and long-term value.”

Skill-based training may be the first ideas that come to mind, but it’s not the only way CIOs can sponsor learning programs. “Skills such as asking good questions, prompting, understanding hallucinations, and critical thinking all need honing,” ThoughtSpot’s Howson says.

Recommendation: Creating a company culture that evangelizes lifelong learning will be critical as AI’s capabilities improve. Leaders should create ongoing learning opportunities, not just those tied to job, role, or skill transitions, which can be a source of distrust if employees tie them to transformation initiatives.

Focus innovators on AI in customer experiences

Most enterprise SaaS companies are adding agentic AI capabilities to their platforms. HR agents guide managers through performance reviews, supply chain agents can take action during global disruptions, and employee experience agents handle meeting follow-ups. A key objective when evaluating any employee workflow AI agent is whether the enterprise data presented to it leads to smarter decisions and faster actions.

The real innovations to come will involve building AI agents into customer experiences. I expect to see more healthcare agents improving patient experiences, financial service agents aiding banking customers with their investments, and retail agents enhancing the shopping experience.

But successful experiments that lead to production deployments will require increased collaboration between innovation and frontline teams.

“Frontline teams that are actively using the AI also gain a valuable stake in the trajectory of its implementation within their company,” says Ashley Moser, CCO at MelodyArc. “Leaders should create excitement for them to utilize AI, then gather feedback to ensure the AI directly addresses customer needs, since these teams already have direct customer insight.”

Recommendation: Product-based IT organizations have been much better at connecting customer input to the agile development process than waterfall project management. Plan to up the stakes when developing AI agents, as teams will need collaboration from more disciplines and functional roles.

It’s counterintuitive that in introducing AI agents, a key success factor is leading change management efforts to impact and influence people. But AI agents are a technology, and failing to introduce change management practices early and tailored to influencing employee segments is one reason why many AI efforts will fail to deliver business outcomes.