From Mystery to Method: Diagnosing Glaze Flaws with AI

Every potter knows the frustration: a glaze that worked perfectly last month now crawls or blisters, and you’re left guessing—was it the clay, the mixing, or the firing? This troubleshooting rabbit hole consumes precious studio time.

The solution isn't more intuition; it's systematic data analysis. By applying a structured, AI-assisted framework, you can transform flaws from mysteries into solvable problems.

The Core Principle: The Comparative Flaw Matrix

Stop troubleshooting in isolation. The key is to systematically compare the faulty batch against a historically successful control batch. This direct comparison, guided by a predefined "Flaw Matrix," isolates variables and reveals the root cause. Your AI tool's purpose is to execute this correlation search at speed, sifting through your recorded data to find meaningful deviations.

For instance, a tool like Klirr (a studio-management platform with AI insights) can automate this by generating batch consistency reports. You don't ask, "Why is this crawling?" You command: "Compare the material weights, supplier sources, and kiln curve for Batch #024 (crawling) to Batch #018 (successful)."

Mini-Scenario: Your latest translucent glaze is suddenly cloudy. Instead of adjusting everything, you have your system compare all data for this cloudy batch against your last clear batch. The AI highlights a 5% increase in feldspar from a new supplier as the only significant change.

Implementation: Three High-Level Steps

1. Digitize Your Process: Consistently log key data for every batch—exact material weights (including source), mixing conditions (e.g., humidity), and full firing schedule curves. This builds your searchable database.
2. Define Your Flaw Library: Create your "Flaw Matrix"—a simple list linking common flaws (crawling, pinholes, clouding) to their most likely data-related culprits (e.g., crawling -> high surface tension, often from material batch variation or fast ramp rates).
3. Query with Purpose: When a flaw appears, initiate a structured "Correlation Search." Isolate the flaw, reference your matrix for likely causes, and then task your system to compare the faulty batch's data against a control batch across those specific parameters (material sources, kiln curve overlays).

Key Takeaways

Diagnosing flaws shifts from guesswork to methodical investigation by leveraging historical data. The core action is the controlled comparison—contrasting a failed batch against a successful one across precise metrics. Implementing this begins with consistent data logging and a clear framework for what to analyze when specific flaws occur. This AI-augmented approach turns troubleshooting into a repeatable, knowledge-building process.