Why some AI initiatives scale while others stall

Most organizations begin their AI journey by asking the same questions:

• Which model(s) should we use?
• Which use cases should we prioritize?
• How much productivity can we gain?

These are important questions, I ask them too, but they don't address the organizational changes required to translate AI capabilities into sustained business impact. Too many approach AI like a software rollout, when in reality it requires fundamental changes.

After the pilot phase, leaders often run into a gap between AI enthusiasm and AI adoption. A small group of power users finds real value. Everyone else is somewhere on a spectrum from curious to overwhelmed to quietly ignoring the whole thing until someone makes them care. The tools are available, but the way work happens has not really changed.

The organizations generating the most value from AI understand that deployment is only the beginning. Real transformation requires questioning long-standing assumptions:

• Why do we do it this way?
• Which steps are still useful?
• Which ones only exist because they used to be necessary?
• What would this workflow look like if we designed it today, knowing what AI can do now?

Sustained impact requires changes to workflows, governance, and processes. Yet many organizations lack the foundation needed to support those changes, causing AI initiatives to lose momentum before they can scale.

The capability gap: The human factor behind AI success

One of the biggest surprises I have seen during this wave of AI transformation is how uneven AI capability can be inside the same organization.

Two employees can sit on the same team, perform similar jobs, and have completely different relationships with AI. One uses it to summarize documents or draft emails. The other uses it to automate workflows, connect systems, and rethink how work gets done.

Many employees still interact with AI exclusively through a chat window. Some are limited by organizational restrictions or knowledge gaps. Others are trying to do their actual job, keep up with meetings, meet deadlines, and figure out where AI fits without turning it into a second job. Until organizations close that gap, AI will remain concentrated among a small group of power users rather than becoming a catalyst for broader organizational change.

Understanding why that gap exists is key to understanding why some AI initiatives scale while others stall.

The three challenges that cause AI initiatives to stall

1. Could focus on tools before behaviors

   Many organizations assume that purchasing an AI platform and providing basic training will naturally lead to adoption.

   Technology adoption is ultimately a human challenge.

   Employees need time to experiment, learn, fail, and discover where AI can create value in their own work. Without that process, adoption remains surface-level. The technology exists, but it never becomes embedded in daily operations.

   The result is often a small group of enthusiastic users generating value while the broader workforce remains largely unchanged.

2. Productivity gains don't automatically transform workflows

   Even when employees embrace AI, organizations often struggle to achieve meaningful transformation because existing workflows remain the same.

   Many focus on using AI to speed up individual tasks like drafting emails, summarizing documents, or generating content. While those efficiencies matter, they rarely deliver the workforce-wide gains leaders expect.

   The larger opportunity comes from rethinking how work moves through the organization. AI creates the most value when organizations redesign processes around new capabilities rather than simply accelerating existing ones. The difference between a successful AI initiative and a stalled one is often the difference between improving a task and transforming a workflow.

3. Organizations are entering the agent era without a governance model

   The next challenge is already emerging.

   As AI evolves from copilots to autonomous agents, organizations must answer increasingly complex questions about permissions, accountability, compliance, and risk.

   Organizations have spent decades developing governance models for human users. Agents introduce an entirely different set of considerations. They can access information, interact with systems, and execute tasks at a speed and scale humans cannot.

   The challenge is balancing innovation with control, which is not a new challenge. Move too slowly and organizations risk missing opportunities. Move too quickly and they introduce new security risks without adequate guardrails.

   This is especially true as agents begin interacting with enterprise applications, sensitive data, MCP servers, agent skills, and external systems.

Building the right framework to scale AI

Organizations looking to scale AI should focus less on generating hundreds of use cases and more on creating the conditions for adoption.

Start by establishing a clear AI strategy. Align model selection, security requirements, cost considerations, and long-term architecture. Most importantly, define success in terms of workforce outcomes.

Next, invest in workforce readiness. AI boot camps, hands-on training, internal communities, and opportunities for experimentation help employees build confidence, develop practical skills, and discover how AI can transform the way they work.

Then evaluate the technology environment and the state of identity and access management. Legacy systems, identity infrastructure, data sources, and integration points will often determine how effectively AI can be deployed at scale.

As organizations move beyond chat interfaces and toward agents, they also need a strategy for managing access. Applying principles such as least privilege access, privileged access management, and Zero Trust can help ensure agents have access only to the systems, data, and actions required to perform their intended tasks—nothing more. Building these controls early allows organizations to enable innovation without introducing unnecessary risk.

Finally, identify champions who can help others discover valuable use cases, share successes, and accelerate adoption across teams. Once employees understand the technology and the right guardrails are in place, begin reimagining workflows around new possibilities. This is where organizations move beyond individual productivity gains and start unlocking workforce-wide transformation.

Looking ahead

As AI continues to evolve, the conversation will shift from helping people access AI to helping AI access enterprise systems safely.

The organizations that succeed won't necessarily be those with the most advanced models. They'll be the ones that close the capability gap across their workforce, rethink how work flows through the organization, and establish the identity, access, and governance foundations needed to support autonomous agents at scale.

Learn more about securing and optimizing AI agent access in enterprise security environments.