Skip to content
Fast Company Executive Board

The Fast Company Executive Board is a private, fee-based network of influential leaders, experts, executives, and entrepreneurs who share their insights with our audience.

The hidden cost of speed

When AI isn't the whole problem.

[Photo: Getty Images]

Profile photo of Andrew Duncan

Chief Executive Officer · QualityAI

AI is accelerating the way organizations operate, compressing decision cycles and embedding action directly into systems rather than in meetings. Pricing adjusts continuously, customers are routed automatically, and operational responses adapt in real time, often without any explicit moment of approval.

As this shift becomes normalized, many organizations are discovering that while AI is speeding up the business, the business itself is struggling to keep pace.

The underlying tension lies in how organizations are structured, governed, and held accountable, all of which were designed for a world in which decisions occurred intermittently and at a pace that allowed human review.

A change in tempo, not just tooling

For decades, organizational decision-making followed a recognizable cadence. Issues surfaced, then discussed, approved, and acted upon, with governance, assurance, and accountability anchored to those stages.

AI has altered that cadence by turning decisions into continuous execution. After defining objectives and constraints, systems act repeatedly and autonomously, adjusting behavior as conditions change rather than waiting to be revisited.

Most organizations still rely on operating rhythms built around periodic review. Leadership meetings, risk forums, and audits remain the primary mechanisms for oversight, even as decision-making moves into live systems.

This creates a widening gap between how decisions are made and how they are governed.

Speed exposes structural limits

As decision velocity increases, time-dependent control mechanisms begin to strain. Approval gates, manual checks, and handoffs were originally introduced to manage risk and clarify ownership, even if they also introduced delays.

As mechanisms are removed in efficiency’s name, the organization loses the points at which responsibility was exercised and recorded.

Oversight becomes increasingly retrospective. Leaders review outcomes after they occur, often with limited visibility into how those outcomes emerged in the first place. This reflects a mismatch between static control structures and dynamic systems.

Recently, a banking client implemented real-time fraud controls that automatically restrict accounts. While effective, it created internal tension because teams could no longer clearly point to who owned a decision when a customer challenged it.

Scale changes the role of humans

The challenge is compounded by volume. AI-driven systems do not simply act faster but more frequently and across more decisions than human-led processes can reasonably absorb.

Human involvement shifts away from direct decision-making and toward observation and exception handling. One client described how a team that previously reviewed dozens of cases per day now oversees thousands of automated decisions, stepping in only when thresholds are breached.

As humans move further from the decision path, accountability becomes harder to localize. Responsibility still exists, but it is distributed across systems, teams, and configurations rather than attached to individual choices.

When decisions blend into system behavior

At sustained high speed, decisions stop feeling like deliberate acts and begin to register as background system behavior. Actions such as rerouting a customer or adjusting an offer occur without discussion or pause, making them harder to recognize as decisions that may later need to be explained or defended.

In a recent discussion, a client noted that customers were being routed away from live agents by automated logic, yet internally, no one felt they had “made” that decision.

This helps explain why executives increasingly struggle to articulate why a particular outcome occurred, even when systems appear to be functioning correctly. The decision was embedded in logic and executed automatically, leaving little trace of when or how it was made.

Over time, that loss of explanatory confidence undermines trust, even when performance remains stable.

Governance designed for pause struggles with motion

Most governance models assume that decisions await review. Committees, signoffs, and escalation paths are structured around identifiable moments of choice, with the expectation that action follows approval.

Continuous systems invert that relationship, with action unfolding constantly and review occurring later, if at all.

As a result, governance increasingly reflects what has already happened rather than shaping what is about to occur. Policies and guardrails, written at a fixed point in time, struggle to remain aligned with systems that adapt continuously.

Quality as a stabilizing capability

In this environment, quality is being redefined. Rather than serving as a final checkpoint, it is becoming the capability that stabilizes fast-moving systems while they operate. Ongoing signals and behavioral visibility provide leaders with evidence of how systems behave and whether they remain within their intended bounds.

This allows organizations to absorb variability without halting progress, supporting speed while preserving confidence. Quality becomes less about inspection and more about maintaining trust in motion.

Keeping pace without losing control

As AI continues to accelerate business activity, organizations face a choice about how they respond. Slowing systems down rarely delivers a competitive advantage. Redesigning control, accountability, and quality to function at speed offers a more sustainable path.

Andrew Duncan is the CEO of Qualitest.