AI’s Biggest Challenge Isn’t Capability. It’s Organizational Discipline.

Many organizations have embraced AI experimentation but struggle to turn it into consistent business value. This article explores why AI’s biggest challenge isn’t capability but organizational discipline, and how governance, craftsmanship, reuse and quality control help organizations move from experimentation to meaningful outcomes.

Everywhere I look, people are talking about AI adoption becoming a concern. Not because the technology doesn’t work, but because many organizations are still trying to work out how to turn experimentation into measurable outcomes and business value. How do we measure its impact on engineering delivery? How do we know whether it’s helping business teams achieve better results?

To me, that’s an incredibly human challenge, often misunderstood because it sits in such a technical space.

The question isn’t what the tools can do, or even how they’re doing it. We’ve spent the last few years proving it can do a lot. The real question is what people choose to do with it, and whether those choices are making their work better or more productive.

That’s why I think AI’s biggest challenge isn’t capability. It’s organizational discipline.

Moving beyond AI experimentation

Over my 20-plus years in IT helping organizations understand what their challenges are and define where they want to go next, one thing has remained consistent: Getting people into a creative mindset is often the hardest part of any initiative. We’ve developed countless ways to encourage that shift in thinking. Whether through workshops, brainstorming sessions or encouraging teams to experiment with new tools, the goal is the same: Create space for new ideas and challenge established ways of working. That’s exactly how many organizations have approached AI adoption, and that’s been valuable.

But I sometimes wonder if we’ve become stuck in exploration instead of invention.

In any successful workshop, creativity is only the beginning. The best ideas are challenged, refined, tested and improved. The process isn’t simply to create. It’s to create, critique, select, develop and iterate. The same is true of engineering, innovation and craftsmanship. Progress comes not just from generating new ideas, but from evaluating them, improving them and building on what works. Creativity matters, but so does discipline. Without that balance, organizations risk generating more ideas without creating more value.

Why more isn’t always better

But AI is changing that in a lot of scenarios. Perhaps the creativity we’ve all been bringing to it has been released without the discipline of the inventor, craftsman or engineer. We’ve become good at creating, but not always as good at being critical.

With this new technology making creation fast and easy, are we using it to accelerate craftsmanship and quality, or volume and variability alone? 

My fear is the latter, in many cases. Not because AI can’t produce quality outcomes, and not because craftsmanship isn’t still possible. Rather, the previous constraints of time and effort that once forced us to be more selective and deliberate have largely disappeared.

When the effort required to create something new is low, do we take the time to refine, improve and build on what already exists? Or do we start over again from scratch, hoping the next version will be better?

I think most of us recognize that temptation. If a colleague tells us they’ve built a prompt, skill or agent to summarize meeting notes, we might consider using it. But many of us will go away and create our own. And if it doesn’t quite work, we might start again.

For an individual tool, that’s rarely a problem. But across engineering teams, projects and entire organizations, the impact becomes much more significant. The result can be hundreds or thousands of slightly different tools, prompts and workflows, all created to solve similar problems in slightly different ways.

That’s where organizational discipline becomes increasingly important. 

Building better tools

My grandfather was a toolmaker. He made manufacturing machine parts. The margin for error on any individual part was small, but the margin for error on a tool creating those parts was even smaller.

It was also a significant investment to create or amend a tool, so the process was deliberate and disciplined. In many ways, it couldn’t be more different from the world of AI today, where creating a new prompt, agent or workflow can take just minutes.

Yet in both cases, we’re building tools.

That brings us back to the challenge. In a world where we have such power to create at our fingertips, how do we ensure what we create is useful, valuable and of good quality? 

I think it comes down to the tools themselves. The quality of the tools we build often determines the quality of the outputs they produce. Build a culture of craftmanship around tool creation, and you help build a culture of quality outputs. For me, that means bringing discipline back into the process. Create, critique, select, develop and iterate. 

More organizations are beginning to look beyond AI adoption and understand what good practice looks like in reality. They’re asking:

• How do we improve visibility?
• How do we measure value?
• How do we maintain quality and efficiency while encouraging innovation?

These questions will only become more important as AI use and costs, both financial and environmental, expand and organizations’ reliance on them increases. The answer is introducing discipline to this creative process. That can be done by championing tool sharing and iteration, setting expectations around tooling quality, defining best practices, and building in KPI measures and quality controls. 

Bringing discipline back to AI

We’re bringing our engineering expertise to this very human business problem, finding ways to introduce control without stifling innovation, measuring value without overcomplicating the solution, and sharing best practice and tooling across the organization. 

At Ensono, we’re looking at all aspects of this challenge, both for ourselves and for our clients. We have teams of AI specialists taking ideas from across the business and developing them in a controlled environment, carefully evaluating, iterating and governing the tools before making them available more broadly. Some of those solutions proactively recommend maintenance work on clients’ mainframe estates or identify resource optimizations. Others support operational activities, such as report writing or quote generation.

Across our digital and consulting practices, we’re also applying these lessons to software delivery. We’re helping clients embed Spec-Driven Development approaches, establish reusable AI-enabled tooling, and define the operating models, KPIs and best practices needed to support AI adoption at scale. 

Like many technology shifts before it, this challenge ultimately comes down to governance. Not governance in the sense of restricting use, but in creating the conditions for good ideas to be refined, shared, improved and scaled.

AI has made creation faster and easier than ever before. That’s where much of its power lies. But the organizations that realize the greatest value won’t necessarily be the ones creating the most tools, prompts or agents. They’ll be the ones applying the discipline to identify what works, improve it over time and build on it. In the next phase of AI adoption, craftsmanship may matter just as much as creativity.

Frequently Asked Questions

Why are organizations struggling to realize value from AI?

Many organizations have successfully encouraged AI experimentation, but experimentation alone doesn’t guarantee results. The challenge is turning individual prompts, agents and workflows into repeatable, measurable business value. That requires processes for evaluating, refining, sharing and governing what works.

What is organizational discipline in AI?

Organizational discipline is the ability to move beyond experimentation and apply consistent standards to AI adoption. That includes establishing ownership, measuring outcomes, encouraging reuse, defining best practices and creating governance processes that support quality without limiting innovation.

Does AI create a governance challenge?

Yes. AI makes it easier than ever for employees to create new tools, automations and workflows. While that can drive innovation, it can also introduce duplication, inconsistency and varying levels of quality. Governance helps organizations balance creativity with control.

Why isn’t creating more AI tools always better?

The value of AI doesn’t come from the number of tools an organization creates. It comes from the quality, usefulness and adoption of those tools. Organizations often achieve better outcomes by improving and scaling successful solutions rather than continually building new ones from scratch.

How can organizations improve AI quality and consistency?

A good starting point is to treat AI tools the same way they treat other business-critical assets. Encourage teams to share solutions, establish quality standards, define success metrics, and continuously refine what works. The most successful organizations combine experimentation with ongoing evaluation and improvement.