Enterprise AI spending is accelerating, but most organizations still struggle to turn isolated experimentation into repeatable operational behavior.
Companies are rolling out copilots, AI agents, and automation tools at speed while pressuring teams to integrate them immediately into daily work. Yet, in many organizations, employees are still expected to figure out practical AI workflows on their own.
That tension is especially acute for marketing teams, where workflows around search, content, and optimization are changing rapidly. Teams are now expected to increase output while simultaneously rebuilding how work gets done.
Over the past several months, I’ve worked closely with marketing and digital leaders through Optimizely’s Opal University, a hands-on AI training initiative focused on helping teams build practical AI workflows tied directly to their responsibilities.
Nearly 1,700 companies are now using Optimizely Opal, including teams from LinkedIn, Deloitte, EY, Bloomberg, and KPMG. During the program’s first cohort, participants built 375 AI agents tied to recurring operational workflows in just five days.
What stood out most was how similar the organizational bottlenecks were across companies. The companies struggling most rarely lacked access to AI tools. The bigger problem was structural: Employees were learning AI in isolation while organizations treated adoption as an individual skill issue instead of an operational redesign challenge.
Why enterprise AI adoption still breaks down
As we worked with participants during Opal University, several recurring patterns emerged that help explain why many AI initiatives struggle to scale across organizations.
Most of the friction had little to do with access to technology. Instead, the challenges centered around how teams were learning, experimenting with, and integrating AI into everyday work.
1. AI adoption is creating a growing “power user gap”
One of the clearest patterns was the emergence of what many teams described as a “power user gap.” In many organizations, AI capability quickly concentrates within a small group of highly motivated employees while broader teams fall behind.
That imbalance creates scaling problems almost immediately. The employees moving fastest with AI become unofficial internal experts while still managing their normal responsibilities. Meanwhile, AI experimentation stays siloed because teams solving similar problems are not sharing workflows, prompts, or learnings with one another.
Over time, companies risk creating internal AI dependency around a handful of employees instead of building repeatable systems across teams. That makes adoption harder to sustain long term.
2. Employees are being asked to experiment with AI under pressure
Another challenge is that many employees are trying to learn AI while managing growing pressure around productivity, changing workflows and job security. Marketing teams are seeing this firsthand as familiar search, content and optimization tactics are rewritten in real time.
That tension is already reshaping broader marketing workflows. Our recent Passion-Pressure Paradox research found that while 61% of marketers say AI saves them time, only 36% say it meaningfully creates more space for strategic work. Instead, many teams are navigating greater operational complexity alongside rising expectations around output and AI fluency.
Many participants arrived at Opal University already convinced they were behind. Across industries, AI fluency is quickly shifting from a competitive advantage to a baseline expectation — and that perception changes how people engage with AI experimentation.
Employees rarely test new ideas openly when they feel professionally vulnerable. When people are anxious about performance or relevance, they become less likely to share unfinished work, failed experiments or unconventional approaches others could learn from.
3. Many companies are trying to transform too much, too quickly
Many organizations are approaching AI adoption with transformation ambitions that far exceed operational readiness. Teams frequently attempt large-scale workflow automation before employees have mastered smaller operational use cases.
That approach can slow adoption instead of accelerating it. Employees may struggle to connect AI systems to practical day-to-day work while leadership teams become focused on theoretical transformation rather than operational improvement.
Organizations are also accumulating AI tools faster than they are integrating them into actual work. In many cases, the issue is not whether teams have access to AI platforms. The issue is whether employees understand how to operationalize those systems in ways that meaningfully improve their work.
4. Most organizations still haven’t built a shared AI learning environment
Many employees are still learning AI on their own instead of as a team. Even groups tackling similar operational challenges aren’t consistently sharing workflows, prompts, or lessons learned.
Without recurring environments for experimentation and collaboration, AI learning becomes fragmented and difficult to sustain across organizations. Employees may experiment with AI privately, though those learnings rarely become operationalized at the team level.
Many companies still treat AI learning as optional self-development instead of a standard part of operations. Because of this, employees feel they need to build AI skills after hours rather than through structured, leader-supported training.
Marketing Technology News: MarTech Interview with Theresa Pham, Head of Product @ Wayvia
What Opal University revealed about how teams actually adopt AI
While the program surfaced several barriers to adoption, it also revealed clear patterns around what helps teams operationalize AI more effectively.
Participants moved fastest when AI learning was collaborative, tied to specific workflows, and embedded into day-to-day responsibilities.
1. Teams learn faster when AI adoption becomes collaborative
Collaboration continues to play a major role in how marketers experience work itself. According to our research, 40% of marketers identified collaboration and shared team energy as one of the most meaningful parts of their work experience. That same dynamic became especially visible throughout Opal University.
One of the biggest takeaways from Opal University was how quickly people gained confidence once learning became collaborative. Participants openly shared prompts, workflows, and failed experiments throughout the program. That visibility reduced hesitation around AI experimentation and made the learning process feel much less intimidating.
Confidence improved quickly once participants realized most teams were still figuring this out in real time. Many participants entered the program assuming other teams were significantly further ahead with AI adoption. When people began sharing workflows openly, it became clear that most organizations were navigating very similar challenges.
That collaborative environment also accelerated practical adoption. Instead of experimenting in isolation, participants were able to see how peers in similar roles were operationalizing AI inside real workflows tied to day-to-day responsibilities.
2. The most valuable AI use cases were operational, not theoretical
Many of the best outcomes from Opal University came from improving recurring operational work rather than pursuing large-scale automation initiatives.
Participants focused on workflows connected to CRO prioritization, performance benchmarking, reporting, and content operations. Several teams saw substantial efficiency improvements within days. CRO prioritization tasks that previously required several hours were reduced to roughly 30 minutes. Performance benchmarking workflows that once consumed six hours were shortened to approximately 18 minutes.
Those results reinforced an important lesson. Organizations gain momentum when teams focus on getting workflows from 80% to 95% rather than trying to reinvent everything at once. The strongest teams were not chasing fully autonomous systems. They were using AI to reduce operational friction inside existing workflows.
That pattern also reflected a broader shift happening across enterprises. Many organizations are now prioritizing AI operationalization over simply expanding tool access. The conversation is increasingly becoming less about what tools companies own and more about whether teams know how to apply them effectively.
3. Structured time and psychological safety accelerated adoption
Another major takeaway from Opal University was the importance of creating intentional space for AI experimentation. Participants were given dedicated time to test workflows, make mistakes, and refine ideas collaboratively without fear of judgment.
That structure changed how employees approached AI learning. Rather than treating AI experimentation as side work, participants were encouraged to explore workflows together during dedicated sessions.
The experience also reinforced how important leadership support can be in scaling adoption across teams. In many organizations, employees were solving similar workflow problems without discussing them together. Teams moved faster when organizations created recurring opportunities for collaboration, encouraged employees to share experiments openly, and celebrated people for trying new workflows, not just outcomes.
Over time, that structure helped transform AI experimentation into repeatable operational behavior instead of isolated individual effort.
4. AI adoption became more effective when integrated into existing workflows
Participants consistently saw the best results when AI agents were connected directly to existing responsibilities rather than treated as separate innovation exercises.
In many cases, the agents reduced operational drag tied to repetitive tasks and allowed employees to spend more time focused on strategic and creative work.
For example, one SaaS company participating in the program used AI agents to support recurring content production and weekly digest creation that previously consumed an entire day each week. After integrating AI into the workflow, the process was reduced to roughly two hours.
The organizations moving fastest with AI are treating it as operational infrastructure woven into everyday work. They are not isolating experimentation inside innovation teams or expecting employees to figure it out entirely on their own.
Operationalizing AI starts with the people
Opal University reinforced a reality many enterprises are still underestimating: AI adoption does not scale simply because companies buy more tools.
The fastest teams were the ones learning collaboratively, experimenting openly, and applying AI to practical workflows tied directly to their everyday responsibilities. Adoption accelerated when employees could connect AI directly to the work they already owned.
Going forward, the companies seeing the strongest long-term AI returns will not necessarily be the ones deploying the most tools. They will be the ones building systems that help teams learn, share, and operationalize AI together at scale.
About the Author of this Article
Steven Male is Senior Director AI Training and Growth at Optimizely
About Optimizely
Optimizely is an AI-powered digital experience platform (DXP) that helps marketing, digital, and product teams accelerate the entire marketing lifecycle.
