Invisible Threads: How Intelligent, Adaptive Processes Drive AI Success

Processes are leaders' top barriers to change and most immature organizational capability, according to a North Highland-commissioned survey of 500 business leaders.
Organizations often neglect processes in favor of surface-level solutions, making them inflexible and unable to keep pace with innovations, especially AI.
Successful AI implementation requires a two-step approach: "zooming out" to understand the organization's knowledge landscape and "zooming in" to execute using data-driven reinforcement learning.
Reinforcement learning with human-in-the-loop is key to continuous improvement, allowing AI systems to learn from experiences while human experts provide guidance and context.
Zooming out and zooming in uses a combination of data, knowledge, and context, three crucial elements for creating adaptable processes that can leverage AI effectively.

Success in today’s competitive, lightning-fast business landscape is not just about having the most cutting-edge data, technology (AI), or top talent. Our survey of 500 business leaders across the U.S. and U.K. revealed a surprising truth: Processes are leaders' top barriers to change, their most challenging (or complex) problem, and their most immature organizational capability.

Processes are critical because they are the invisible threads that orchestrate the complex interplay of resources, knowledge, and action across organizations. They are the backbone of business operations—the critical bridge connecting people to technology.

But they often take a backseat to more visible investments, and have long been viewed as rigid, fixed structures disconnected from actual employee ways of working.

This perception has led to a common pitfall: Rather than determining how processes could be reimagined to drive growth, executive teams neglect them and instead focus on immediate, surface-level solutions—be it modern technology, like fragmented commercial off-the-shelf software (COTS), business intelligence (BI), AI, or workforce changes. Over time, this neglect and narrow focus becomes a self-fulfilling prophecy: Processes fail to evolve with the business, resulting in inflexible procedures that are not equipped to keep pace with rapid innovations.

This age-old challenge is exacerbated by the modern demands of AI. Far too often, leadership teams fall victim to the hype surrounding the next shiny object, rushing to implement new AI technologies without carefully considering how they will integrate with existing processes and ways of working. In their haste to stay ahead of the curve, they fail to ask the crucial question: how can this technology transform alongside our processes?

This shortsighted approach leads to inflexible, brittle processes that cannot withstand the speed and complexity of AI advancements. As a result, organizations find themselves grappling with suboptimal outcomes and squandered value.

But there is another way: To successfully leverage the potential of any innovative technology—including AI—you must prioritize your processes and their continuous improvement and adaptability. By ensuring that your processes are built to evolve alongside new technologies, you can effectively harness their power to drive your organization forward.

In our experience, this process-centric approach involves two key steps—zooming out, then zooming in—and leverages the powerful trio of knowledge, context, and data. While we'll focus on AI in this piece, these principles apply broadly to the integration of any transformative technology:

Zooming out to connect the dots. The first phase of AI implementation is about zooming out and looking at the big picture of your business. Instead of focusing on a narrowly defined AI initiative or use case, you need to consider dependencies and the impact AI will have on your entire organization—from your people, teams, and departments to your processes and ways of working. Start by gathering comprehensive knowledge about your business operations—this means collecting data about your processes, ways of working, technology, people, and everything in between. This knowledge gives you context for how these critical elements of your business can work together to produce the best results. Armed with this insight, you're positioned to identify the most strategic opportunities for AI integration. It empowers you to boldly reimagine your processes, leveraging cutting-edge technology not just as an add-on, but as a core driver of your business objectives.
Zooming in to execute. The second phase uses reinforcement learning to refine your AI processes over time. Reinforcement learning is a data-driven method where AI learns by continuously interacting with its environment. Every action and outcome generates data, creating a feedback loop that guides AI's learning process. The feedback loop monitors how your processes are performing in real-time, keeping an eye on hiccups and inefficiencies and identifying how people (your employees and customers) are using AI tools. With this data, AI optimizes its decision-making and processes over time, learning from experience just like humans do. Crucially, you need to bring in human expertise (which we’ll refer to as human-in-the-loop) to add essential context and knowledge to your data, creating a dynamic learning environment where real-world information and human insights collaborate to enhance the AI system continuously. The result? Processes that evolve automatically, systems that adapt to user needs, quick identification and implementation of new ideas, and the ability to jump on new opportunities faster than ever.

Defining data, knowledge, and context

Before we dive into zooming out and then zooming in, it’s helpful to paint a familiar picture of how data, knowledge, and context work together in AI systems. We can look to self-driving cars, like a Tesla:

Data. Think of data as all the little pieces of information the car collects. It's like it has eyes (cameras) and ears (sensors) that constantly gather facts. For example: There is a red light ahead. The car goes 30 miles per hour. There is an object 10 feet in front of the car.
Context. Context is how the car understands what all this data means together. It's like putting puzzle pieces together to see the big picture. For instance, the red light means the car should stop. It is raining, so it turns on the windshield wipers.
Knowledge. The AI learns and adapts its processes and decision-making over time, based on the knowledge it acquires through experiences and patterns. For instance, a GPS navigation system may initially recommend highways, which are typically the shortest route. But if the driver consistently chooses backroads instead, the AI learns from this behavior and adapts its recommendations accordingly. This adaptability is made possible by the AI's ability to change its default process based on the knowledge it gains from the driver's preferences. By leveraging data and context to build knowledge, the AI can optimize its processes to better serve the driver's needs.

Just like a Tesla, when you have adaptable processes in your business, your systems can learn, evolve, and optimize based on data, context, and knowledge. As a result, your organization can deliver more efficient, personalized, and user-centric solutions.

Let’s get into the two-phase approach that will help you create adaptable, AI-enabled processes.

Phase 1: Zooming out to connect the dots

Taking a step back and adopting a strategic, knowledge-driven approach.

This is the key to ensuring your processes are adaptable and built for the age of AI. It means taking the time to understand your organization's collective knowledge before rushing into AI integration.

For example, imagine Woven, a hypothetical e-commerce company planning to implement an AI-driven product recommendation system.

To avoid pitfalls of AI adoption, Woven partnered with North Highland to zoom out and understand its existing knowledge base. Using advanced analytics tools, it mapped processes, workflows, and existing knowledge.

When we talk about knowledge, we're referring to explicit, implicit, and tacit information that provide context for decision-making. For Woven, this might include:

Explicit knowledge: Documented customer demographics and purchase history
Implicit knowledge: Unwritten rules about seasonal trends and promotional strategies
Tacit knowledge: Employees' intuition about customer preferences, behavior, and trends

Effective knowledge gathering and management ensures that your AI systems have access to reliable, accurate, and contextually relevant information. By understanding how work gets done in your organization, you can identify opportunities for AI to drive value, connect the dots, and make informed decisions about new ways of working that optimize organizational value and realize the potential of AI.

In our Woven example, this might involve mapping out the current product recommendation process, understanding how sales teams currently make suggestions, and identifying what data is already being collected and used.

As David Johnson, North Highland Vice President and Global AI Practice Leader, puts it...

"Knowledge is the manufacturing process that runs on top of data to turn it into a beautiful product. That manufacturing process involves understanding, interpreting, and decision-making based on the different types of knowledge."

Engaging an experienced partner to establish an AI Center of Excellence (CoE ) can be a game-changer in developing a robust knowledge management strategy. A CoE can help you:

Map your knowledge landscape. It’s critical to conduct a thorough assessment of your organization's collective wisdom to identify expertise, unwritten rules, critical processes, practices, and behaviors that drive your operations. This goes beyond mere documentation; it's about uncovering the layers of tacit knowledge that make your organization unique to understand where AI and automation can be integrated to drive productivity and efficiency gains. This comprehensive "atlas" of your organizational knowledge illustrates processes and decision-making pathways, collaboration networks, and innovation hubs. For Woven, this might involve using advanced analytics tools to document the expertise of long-time sales staff, understand the unwritten rules of customer engagement, and map out the decision-making processes for product placement and promotion.
Plan for growth. With your knowledge landscape mapped, you can assess current workforce skills and identify gaps. These gaps represent opportunities for growth, development, and role-based change. Woven, for instance, might spot gaps in data science skills and develop plans to upskill current employees or hire new talent with expertise in AI and machine learning.
Reimagine your processes to amplify employee strengths. A CoE partner can help you examine your knowledge ecosystem and reimagine your processes to ensure you’re creating a symbiotic relationship between human expertise and AI capabilities, enhancing overall organizational performance. This partner can also support employee education to bolster adoption. In Woven’s case, the CoE could help it identify high-impact AI initiatives, develop a scalable strategy, consider potential impacts on people, and redesign product recommendation processes to combine AI insights with human expertise, creating a more effective hybrid approach.
Promote cross-functional collaboration. Prioritizing cross-functional collaboration when creating a knowledge ecosystem is essential for unlocking your organization’s innovative potential. By bringing together diverse perspectives and expertise, you can enrich context and knowledge for better problem-solving; accelerate the identification of creative solutions; foster the development of new skills among team members; align strategic objectives across the organization; and create a more cohesive and unified strategic vision. For Woven, this could involve bringing together the sales, marketing, IT, and customer service teams to provide diverse perspectives on how an AI recommendation system could be most effectively implemented and used.

As you embark on your AI-powered transformation journey, remember that your organization's unique knowledge base is its most valuable asset. For Woven, this knowledge base—encompassing customer behavior patterns, product lifecycle insights, and market trends—will be crucial in training and fine-tuning its AI recommendation system.

By strategically mapping and leveraging your collective wisdom, you create a solid foundation for AI integration that enhances (rather than replaces) human capabilities, accelerates AI development and deployment, and unlocks unprecedented levels of efficiency, innovation, and competitive advantage.

It positions you to inject life into stale processes and transition to future ways of working that enable employees to do their jobs and better understand where they fit into the organization (most of which can be enabled by AI).

While 'zooming out' provides the big picture, the next step is to 'zoom in' and put this knowledge into action. This involves implementing a data-driven approach to continuously refine and enhance your processes, which for our friends at Woven, would mean constantly updating and improving their AI recommendation system based on new data and insights.

Phase 2: Zooming in to execute

Leveraging data for continous improvement.

Now that you’ve mapped your organizational knowledge landscape, it's time to leverage your daily data streams to continuously refine and enhance your processes. Your knowledge map functions like an advanced organizational GPS, showing your current position and available resources. However, to navigate the rapidly changing business landscape, you need constant updates—akin to real-time traffic information for your GPS. This is where reinforcement learning with human-in-the-loop becomes essential.

Reinforcement learning is an AI method where a system learns through trial and error based on feedback from its actions. For example, Woven could use reinforcement learning to optimize its product recommendation system:

The AI recommends products to customers browsing the website. Each recommendation is an "action" in RL terms and the "environment" is the e-commerce platform and customer behavior.
The system receives rewards based on customer actions (clicks, purchases, or ignoring recommendations).
Over time, the AI learns which types of recommendations are more likely to result in clicks and purchases for different customer segments, product categories, or times of day.

This continuous stream of data creates a feedback loop that works to continuously improve the system's performance. The loop applies organizational knowledge (mapped in Phase 1) to contextualize the data and make it actionable.

However, human involvement is crucial. Human-in-the-loop enhances reinforcement learning by providing additional knowledge and context that AI may not be able to learn on its own. In the Woven example, employees could enhance the process by:

Defining the reward structure (how feedback is given to the system) based on business goals and an understanding of customer behaviors.
Identifying relevant customer features or attributes (age, browsing history, etc.) that the AI should consider.
Providing context around seasonal trends, marketing campaigns, or cultural factors that might influence buying behavior.
Ensuring ethical oversight.
Interpreting complex feedback.
Unpacking unusual patterns the AI identifies, providing context (e.g., a sudden change in behavior might be due to a viral social media post).
Periodically reviewing the AI's learned strategies and adjusting based on broader business knowledge.
Providing context for new products that the AI hasn't had time to learn about.
Incorporating feedback from customer service teams to improve recommendations.
Assessing the AI's performance not just on metrics, but on alignment with overall business strategy.

This combination of reinforcement learning and human expertise transforms your approach to AI integration and process optimization. It creates a cycle where data flows between AI and human experts, enabling real-time improvements. As changes are implemented, new data is collected, and the process repeats, allowing your processes and people to evolve together.

The result is heightened agility when responding to market shifts and internal changes while keeping your workforce engaged. A trusted partner can help measure adoption, ensure efficacy, and keep your team aligned with technological advancements, so no one gets left behind.

Your next steps: Transforming your business with adaptable, AI-enabled processes

As you embark on your AI journey, think beyond the hype and focus on the foundation: your processes. By embracing this two-phase approach of zooming out to understand your organizational knowledge and zooming in to implement data-driven, adaptive processes, you're not just preparing for AI—you're revolutionizing your entire business model along the way.

Imagine what your business could achieve if:

Your processes evolved in real-time, adapting to market changes and customer needs.
AI and your employees worked in perfect harmony, each amplifying the other's strengths.
Innovation was not only encouraged—it was woven into the very fabric of your operations.
Employees were empowered, engaged, and constantly growing alongside your technology.

This isn't a far-off dream. It's the tangible result of leveraging data, context, and knowledge to unlock the potential of AI and make your processes the hero of your transformation story.