For the past several years, artificial intelligence has been everywhere in enterprise conversations and nowhere in actual results. Most organizations have experimented, many have piloted, but very few have operationalized AI in a way that meaningfully moves the needle. In fact, the numbers tell a sobering story: the vast majority of enterprise AI initiatives fail to deliver tangible business value, with estimates suggesting as many as 80 to 95% fall short of expectations.
For enterprise leaders, this gap between promise and performance isn’t just frustrating, it is costly. It also raises an important question: why has AI struggled to deliver despite the explosion of internal and external data, and the significant technology investments made to support enterprise processes and teams?
The answer lies less in the technology itself and more in how it’s being applied across real-world, interconnected operations. Most organizations are not failing because AI doesn’t work, but because of fragmented data, siloed processes, and the challenge of integrating AI into existing systems and workflows.
Why AI Struggles in Real-World Operations
If the first wave of AI was about prediction, the next wave is about action. Leading organizations are moving beyond static models toward systems that can sense, analyze, decide, and act in near real time, often described as agentic capabilities. These systems do not just generate insights; they participate directly in execution.
We are already seeing what this looks like in practice. In fast-moving consumer sectors, AI-driven approaches are compressing product development cycles from months to days by linking demand signals directly to design and supply decisions. In retail and food service, digital twin environments are enabling teams to simulate disruptions and resolve them in minutes instead of days, reducing manual effort while improving service levels and inventory efficiency.
The common thread is the integration of AI into the operational fabric of the business. Four characteristics stand out:
A shared, semantic understanding of the business – Enterprise intelligence is built on a unified representation of the business across structured and unstructured data, enabling AI to understand how information, decisions, and outcomes are connected.
Workflow-driven simulation and execution – AI must operate within real business processes, simulating outcomes across workflows and applying business logic to guide decisions from planning through execution.
Orchestrated decision-making across the enterprise – An orchestration layer connects people, data, and systems, enabling more collaborative, informed decisions and ensuring those decisions can be executed across existing systems of record.
Continuous learning and improvement – AI systems must continuously learn from outcomes, improving decision quality over time and adapting to changing conditions.
For all its sophistication, AI must also feel simple to the user. The underlying complexity does not disappear, but it must be abstracted. The real test is whether organizations can deliver accurate, real-time answers without exposing the complexity that makes them possible.
Increasingly, this also means making these capabilities accessible beyond technical teams, through self-service environments that allow business users to define, test, and adapt decision processes without relying entirely on specialized development resources.
Why Supply Chains Are the Proving Ground for Enterprise AI
Supply chains reflect the operational “physics” of the enterprise and are among its most consequential domains. Every decision, what to make, where to ship, how much to hold, has direct financial and customer impact. That complexity, when fragmented and managed in silos, is precisely why AI has struggled here. It is also why success in supply chains matters more than anywhere else.
But the implications go beyond supply chains alone. The highest-value opportunities are centered on solving these challenges because they are foundational to how enterprises operate. Supply chains are simply the most demanding proving ground for a broader shift toward real-time, connected decision-making across the business.
We are now at an inflection point. Advances in data architecture, cloud scalability, and AI techniques are converging to make enterprise-grade deployment practical. At the same time, market momentum is accelerating, with investment in enterprise AI outpacing traditional software categories and signaling a long-term shift in how organizations operate. For leaders, this creates both urgency and opportunity.
The lesson from the past few years is clear: isolated AI projects do not scale. Incremental experimentation alone will not get organizations where they need to go. What is required instead is a system-level approach that connects data, decisions, and execution across the end-to-end enterprise. This is where orchestration becomes critical.
Orchestration is not about replacing planning. It is about elevating it by connecting planning with execution, aligning decisions across functions, and ensuring that AI operates within the full context of the business. It enables organizations to move from reacting to disruptions to proactively managing them.
From a technology perspective, this means building environments where data, AI models, business rules, and human expertise coexist, and where decisions can be simulated, tested, and executed in a continuous loop.
From an organizational perspective, it requires rethinking how teams work. The most successful companies are investing in cross-functional capabilities, bringing together domain experts, data scientists, and operators to translate AI potential into operational reality. What’s emerging is not just a new set of tools, but a new operating layer that connects decisions across the enterprise.
From AI Experimentation to Enterprise Impact
AI in enterprise is no longer a question of “if.” It is a question of “how.” The hype cycle is giving way to a more pragmatic phase focused on measurable outcomes, operational integration, and sustained value creation.
Organizations that continue to treat AI as a series of experiments will fall behind, while those that embed it into the core of how they operate will gain a meaningful competitive edge.
McKinsey’s research reinforces this divide. High performers are nearly three times more likely than others to redesign workflows around AI, rather than simply layering it onto existing processes, and are significantly more likely to view AI as a driver of enterprise-wide transformation.
The path forward is not about chasing the next breakthrough model. It is about building the
foundation for continuous, connected decision-making and execution at scale. In complex operations, value does not come from knowing more. It comes from consuming data, contextualizing it, analyzing it, making decisions, and acting on them.
By Manik Sharma, Chief of Agentic Solutions, Kinaxis
Manik is a seasoned supply chain and digital transformation leader, bringing more than 30 years of experience driving operational and organizational change across industries and global markets. He is known for building and scaling go-to-market strategies, accelerating sales momentum, and delivering measurable customer value through strong cross-functional orchestration.
Currently Chief of Agentic Solutions at Kinaxis, he leads the development and execution of AI-driven strategies that help organizations transform decision-making and operational performance. His career spans leadership roles at Celonis, Palantir, Coupa, and previously Kinaxis, where he has consistently driven growth, innovation, and customer impact.
His expertise includes digital transformation, supply chain management, advanced analytics, and enterprise strategy, with a focus on helping organizations adapt and compete in increasingly complex environments.
The post From AI Experiments to Operational Impact: What It Really Takes for Enterprises to Realize Value appeared first on Logistics Viewpoints.