From Guesswork to Glaze Code: AI for the Small-Batch Potter

Ever pulled a perfect piece from the kiln, only to find its twin in the next firing is a dud? For small-batch ceramic artists, this inconsistency is the ultimate frustration. Replicating success relies on a fragile memory of countless variables. It's time to move from intuition to data.

The Principle: Treat Every Firing as a Dataset

The core idea is to stop viewing firings as singular events and start treating them as entries in a searchable database. Your goal isn't just a beautiful pot, but a complete record of the conditions that created it. This data is the fuel for AI automation.

By meticulously logging both Descriptive Data (what happened) and Prescriptiv e Data (what you’ll change), you build a corpus an AI can analyze. For instance, note that "Glaze X always works with a 15-minute soak in my kiln" or that your "bottom shelf consistently under-fires by a half-cone." These are not just observations; they are critical data points for pattern recognition.

The Tool: Your Digital Kiln Controller

Your most immediate tool is your digital kiln controller, if you have one. Its data log provides the "Actual Peak Temp & Time," a non-negotiable metric. This objective record replaces subjective guesswork about what "Cone 10" truly looked like in your kiln on that day. It becomes the anchor for all other observations.

Mini-Scenario: You notice a glaze matures perfectly only on the top shelf. An AI tool, trained on your past firing logs that note shelf position and results, could flag this spatial inconsistency and suggest a schedule adjustment for lower shelves.

Implementation: Three Steps to Start

1. Structure Your Logging: Create a standard template for every firing. It must include a unique Firing ID, the Program/Firing Schedule, Kiln Used, Atmosphere notes, and crucially, the Goal versus the Descriptive Data outcome.
2. Digitize the Historical: Input past firing notes—both successes and failures—into a structured format. Even simple spreadsheets work to start. The key is consistency: always log the same variables.
3. Apply Basic Analysis First: Before complex AI, use your data. Filter past firings by Goal or Problem. Can you see a pattern linking "Kiln Won’t Reach Temperature" to specific weather or an "Old Assumption" about clay body preparation?

By building this disciplined, data-centric practice, you lay the groundwork for AI to eventually automate glaze recipe adjustments and predict outcomes. The kiln's code isn't just temperature and time; it's the entire symphony of variables you can now conduct with precision. Start logging, and turn your process from an art into a reproducible science.