- Scaling AI can expose workflow friction that has become embedded in how an organization operates.
- Removing unnecessary complexity helps organizations improve efficiency at scale and strengthen their competitive advantage.
- As AI usage grows, leaders must address both operational friction and infrastructure costs, including the cost of AI inference.
Table of Contents
At most companies, inefficiencies often build slowly. A workaround solves an immediate problem, an extra approval adds a level of control and a handoff becomes part of the process. Over time, those decisions can compound, making it harder for work to move efficiently through an organization.
AI is starting to make that friction easier to see. As enterprises move from experimenting with AI to implementing it more broadly, existing bottlenecks become harder to ignore. Processes designed around manual work may no longer make sense when AI changes how that work gets done.
This creates an opportunity that goes beyond making existing processes faster. Leaders can ask a deeper question: What is actually necessary for the work to get done?
We believe the competitive advantage does not come only from scaling AI faster. It comes also from making it easier for work to flow in the first place.
That requires looking for ways to reduce inefficiencies on an ongoing basis. As AI scales, organizations will need to keep looking for sources of friction and determine whether the systems around the technology are helping it create value or are simply getting in the way.
Where AI Exposes Workflow Friction
As AI becomes more integrated into an organization, the sources of friction it exposes can take different forms. Some bottlenecks are obvious. Others have become so embedded in day-to-day operations that they are accepted as part of how work gets done.
Four areas are particularly important:
- Knowledge Friction: Information loses value when employees have to search throughout disconnected systems or rely on other teams to find what they need. AI offers a way to make institutional knowledge easier to access while revealing where fragmented systems get in the way.
- Execution Friction: Manual steps can slow work down as it moves through an organization. Agentic AI offers a way to automate parts of that execution while giving leaders an opportunity to reconsider which steps are still necessary.
- Time-to-Market Friction: AI helps organizations identify where work stalls between ideation and execution, creating opportunities to simplify processes and respond more quickly to changing market conditions.
- Security and Compliance Friction: Greater automation may improve oversight, but it also introduces new risks. Addressing these considerations as AI systems are built helps keep governance from becoming an afterthought.
We believe that using enterprise AI to create value through these operational layers involves more than improving specific workflows. The larger opportunity is to make the organization itself easier to scale. As unnecessary complexity is removed, efficiencies can spread throughout the organization and create a compounding competitive advantage.
How Operational Efficiency Compounds at Scale
The benefits of removing friction often reach beyond a single workflow. As AI becomes more embedded in an organization, new bottlenecks emerge and existing processes may need to evolve with the technology.
Lower latency helps AI-enabled workflows respond more rapidly. Simplifying processes allows work to move with less intervention. Over time, those improvements build on one another, making it easier for an organization to apply what it learns in one workflow elsewhere in the business.
This is what makes efficiency at scale different from improving single tasks in isolation. The goal is to make individual tasks faster, but also create an organization that continues to prioritize the removal of friction as its technology and operations evolve.
But organizational complexity is only one constraint on that advantage. As AI usage grows, the economics of the underlying technology matter too. As an AI system becomes more useful, rising operating costs may introduce a different kind of friction.
The Importance of Understanding the Cost of AI Inference
For companies scaling AI, managing the ongoing cost of running these systems is an important part of making that growth sustainable.
One important variable cost associated with running AI comes from inference, the process through which a trained model uses new inputs to generate outputs. As AI usage increases, understanding how inference costs change at scale becomes increasingly important to the economics of implementation.
As Lip-Bu Tan, CEO of Intel, discussed at Vista Equity Partners’ 2026 Annual General Meeting, “The whole semiconductor foundation and infrastructure needs to massively scale for Agentic AI. The training is pretty much set, but the part I think is much bigger is agentic inference and that has profound impact across every layer.”
The same principle applies here: scaling AI is only part of the equation. Leaders also need to understand what may reduce efficiency as it grows. Whether the friction comes from a workflow or the infrastructure behind it, removing these constraints helps AI deliver greater value over time.
Stay up to date with the latest from Robert F. Smith by following him on LinkedIn and subscribing to his YouTube channel.