Solving the Engineering Problem that Makes AI Actually Useful: Building the Axle

2026-01-05 · Source: Modern Data 101 · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Cloud Computing & IT Infrastructure · Depth: Advanced, long

Summary

This article, "The Wheel and the Algorithm | Part 2" by Sagar Paul, extends an analogy comparing AI adoption to the historical development of the wheel. It argues that while AI models are the visible "wheel," the true engineering challenge lies in developing the "axle"—the precision integration layer connecting AI to enterprise data and workflows. A 2024 computational mechanics study in Royal Society Open Science found the wheel-and-axle system evolved over 500 years, requiring simultaneous solutions to challenges like roundness, tolerance, strength, and lubrication. Similarly, enterprise AI faces four "tolerance challenges" simultaneously: format, quality, temporal, and semantic. The article proposes "data products" as the solution, acting as a precision-engineered connection layer that embeds context, quality, and temporal awareness, enabling rapid iteration and breaking historical patterns of slow technology adoption.

Key takeaway

CTOs and VPs of Engineering grappling with AI adoption should prioritize developing a "data product" discipline to address the four critical tolerance challenges (format, quality, temporal, semantic) simultaneously. This approach, akin to precision engineering the "axle" for AI "wheels," will enable faster iteration and clearer line of sight from investment to business value, allowing your organization to leapfrog competitors still stuck in multi-year transformation programs.

Key insights

AI adoption requires solving complex data integration "axle problems" simultaneously, not just developing impressive "wheel" models.

Principles

• Precision engineering is key to AI utility.
• Simultaneous problem-solving is critical for integration.
• Data products enable rapid iteration.

Method

Implement data products to embed format, quality, temporal, and semantic tolerances directly, enabling rapid iteration through a deploy-observe-fix-redeploy cycle.

In practice

• Define quality relative to specific outcomes.
• Treat freshness as a first-class contract.
• Localize change with data products.

Topics

• Enterprise AI Integration
• Data Products
• Data Operating System
• Data Tolerance Challenges
• Iterative AI Deployment

Best for: CTO, VP of Engineering/Data, AI Architect, Data Engineer, MLOps Engineer, Director of AI/ML

Related on AIssential

• The Simplicity Paradox: Why Modern AI Still Depends on Classic Data Engineering — Modern AI's perceived simplicity masks its deep reliance on classic data engineering for robust operation.
• The AI Interview: Julien Moutte, Bentley Systems — AI agents can act as a force multiplier in infrastructure engineering, enhancing efficiency and collaboration without replacing human oversight.
• Small Wins, Big Ripples: The Joy of Building Micro AI Data Products — The article challenges the prevailing "Big AI" narrative of massive overhauls, instead advocating for "Micro AI" data products.
• Why AI Integration Is More About Coordination Than Replacement — AI success hinges on coordinating existing systems and workflows, not rebuilding products from scratch.
• The Hardware Bottleneck AI Can’t Fix — Hardware engineering's lack of software-like tooling creates a bottleneck, hindering AI's impact and requiring specialized data platforms.
• Bridging the operational AI gap — Integrated data and systems are critical for successful enterprise-wide AI adoption and operationalization.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Modern Data 101.