Why Most Manufacturing AI Initiatives Fail Before They Scale

Interview by Nishkam Batta, Founder, HonestAI by GrayCyan

Artificial intelligence has become one of the top priorities for manufacturers all over the world. Yet despite increasing investment, many organizations are still struggling to move beyond isolated pilots and disconnected experiments.

AI vendors continue to tout productivity gains and operational transformation, but the reality inside many manufacturing organizations is far more complex. The real challenge is not choosing the right AI platform, it's preparing the business to successfully adopt AI.

In an exclusive conversation with Nishkam Batta, Nicole Hilgenkamp, former AI Enablement Leader at Cargill and now a manufacturing consultant, shared firsthand lessons from leading AI adoption across a global procurement organization supporting more than 1,100 employees and $14 billion in spend. Rather than focusing on technology, Nicole explained why strategy, data, governance, and culture determine whether AI succeeds or fails.

AI Doesn't Fail Because of Technology

The biggest misconception about manufacturing AI is that companies need better models or better software.

One of the biggest myths around manufacturing AI is that companies need better models or better software.

Nicole has the opposite opinion.

The first challenge organizations face is not choosing AI tools, but realizing that their existing business processes are not prepared for automation.

In one procurement , the team experimented with automating spot buying through generative AI. The project quickly revealed inconsistent product descriptions, incomplete specifications, and fragmented knowledge management.

The organization killed the pilot instead of scaling it. Many organizations begin their AI journey with the purchase of enterprise licenses.

Nicole went a different way.

Her team started with training staff, running hackathons, holding AI office hours and promoting experimentation.

The lesson was simple:

AI only scales when processes, data, and knowledge are standardized first.

Start with People, Not Platforms

Many organizations begin AI adoption by purchasing enterprise licenses.

Nicole took a different approach.

Her team first trained employees, organized hackathons, hosted AI office hours, and encouraged experimentation.

Out of 388 participating employees, the team identified approximately 30 meaningful AI use cases before deciding which opportunities deserved investment. This allowed AI ideas to emerge from employees who understood the work best instead of being dictated solely from management.

Data Is the Foundation of AI

One of the biggest takeaways from the interview is that AI doesn't fix underlying business problems, it amplifies them.

If processes are inconsistent, data is unreliable, or knowledge is scattered across the organization, AI will simply scale those inefficiencies rather than eliminate them. That's why manufacturers must first establish a strong foundation with standardized processes, high-quality data, and effective knowledge management before expecting AI to deliver meaningful business value.

Nicole described her team's move to build standardized process documentation, structured knowledge management and better data governance before trying to deploy AI at enterprise scale.

Culture Is the Real Bottleneck

Throughout the interview, Nicole returned to one recurring theme:

Manufacturers tend to expect workers to embrace AI "right away." Adoption is really a step-by-step thing. Small wins build trust.

Instead of asking seasoned operators to change the way they work completely, organizations should look for frustrations in the daily work and show how AI takes away repetitive work and allows people to focus on higher-value work.

Strategy Before AI

Nicole also challenged one of the most common mistakes manufacturers make: Starting with AI instead of starting with business strategy.

Her recommendation was clear.

Begin by identifying the company's most important strategic objective.

Only then should manufacturers ask where AI can create measurable value.

Every AI investment should directly support competitive advantage rather than becoming another isolated technology experiment.

Give Everyone Their Own AI Assistant

One of Nicole's most practical recommendations was surprisingly simple.

Instead of asking employees to replace how they work, provide everyone with an AI assistant. No matter it's an operator, HR professional, engineer, or executive, every employee has repetitive administrative work.

AI should begin by handling those repetitive tasks. Employees remain responsible for judgment, expertise, and decision-making while AI handles documentation, summaries, reports, scheduling, and other routine activities.

This approach makes AI less threatening and far easier to adopt across manufacturing organizations.

Key Lessons for Manufacturers

Manufacturers looking to scale AI successfully should:

• Start with strategic business priorities, not AI tools.
• Build clean data, standardized processes, and strong knowledge management.
• Train employees before deploying technology.
• Create governance before experimentation grows.
• Focus on solving everyday employee problems to build trust.
• Scale successful pilots only after measurable business value has been demonstrated.

Conclusion

Artificial intelligence is changing manufacturing, but technology alone won't be enough.

The manufacturers that will win are those who invest in people, processes, governance and culture as they do in AI platforms.

The takeaway from Nicole's experience is that successful AI adoption is not about more AI, it's about preparing the organization to use it effectively. Only then will AI move from experimentation to a sustainable competitive advantage.

About the Interviewer