Foreword

To further AI understanding and adoption, Lazarus AI is producing a series of articles titled Artificial Intelligence – Insights for Insurance (“AI-II”). This article addresses a common assumption in AI adoption: that data must be fully cleaned and standardized before AI can be applied. While this holds true in some contexts, it can delay or prevent meaningful progress in others.

‍

The Problem

For fields where analysis and predictive AI are central to a solution, high-quality data is essential. Poor data leads to poor outcomes, often at significant cost. As a result, “clean your data first” has become a widely accepted principle.

The issue arises when this principle is applied universally.

Operational processes rarely have perfect data. Despite significant investment in data improvement, most organizations do not achieve—and will never achieve—perfect operational data. Yet they continue to function despite inconsistencies, gaps, and human-introduced variability. If perfect data were required for execution, many business processes would not operate at all.

‍

The Misapplication

The belief that AI must wait for pristine data creates a bottleneck. It delays adoption, limits the ability to realize value, and puts organizations at risk for falling behind.

In practice, many operational use cases do not require perfect data to benefit from AI. These processes already rely on human judgment to interpret incomplete or inconsistent information. AI can support or augment that same process.

In some cases, AI can also help improve data quality directly. For example, interpreting handwritten inputs or resolving inconsistencies across documents are tasks well suited to AI systems.

The principle is not to ignore data quality, but to recognize that progress does not require perfection.

‍

The Right Approach

Leaders should distinguish between analytical and operational AI use cases.

Where precision and prediction are central, data quality remains a prerequisite. In these cases, AI can become part of the solution: helping to untangle data challenges, rather than being blocked by them. For processes already operating with imperfect data, AI can be applied immediately to improve efficiency and consistency.

Waiting for perfect data introduces its own risk: falling behind and leaving value on the table.

‍

Summary

Data quality matters. But requiring perfection before action is not realistic for most operational environments. Organizations that apply AI selectively, based on where it can add value today, will move faster and learn more quickly. Those that wait for ideal conditions may never reach them.

‍

About Lazarus AI

Lazarus AI develops enterprise-grade AI systems for the insurance industry, public sector, and beyond. Our Applied Intelligence Engine (AIE) enables organizations to eliminate their processing bottlenecks and provides rapid time to value, allowing our customers to compete more effectively with reduced cost, lower risk, and greater speed.

‍