For the ceramic artist, nothing is more frustrating than a perfect glaze recipe failing in the next firing. You meticulously mixed, applied, and fired, only to find crawling, inconsistent color, or under-fired shelves. Replication feels like alchemy, dependent on memory and scribbled kiln logs. You’re not just an artist; you’re a materials scientist running a one-person lab.
Your New Framework: Descriptive vs. Prescriptive Logging
The core principle for mastering your process is separating Descriptive Data (what actually happened) from Prescriptive Data (what you planned to happen). Human memory conflates them. AI thrives on their distinction.
Descriptive data is the raw, observational truth of the firing: "My bottom shelf under-fired by a half-cone," or "Atmosphere: heavy reduction starting at Cone 012, noted by orange flame at peephole." It includes the Actual Peak Temp & Time from your controller log and critical observations like clay body changes. Prescriptive data is your intention: the Program/Firing Schedule (e.g., "Slow Glaze to Cone 6, 10-min hold") and the Goal of the firing.
Mini-scenario: Your famous Celadon turns muddy. Old assumption: "The glaze batch was wrong." New method: Your descriptive log shows a lighter reduction atmosphere noted that day. The problem wasn't the recipe, but a prescriptive variable you didn't track.
Implementing Your Digital Kiln Log
Step 1: Structure Your Firing Log. Create a digital record for every firing. Use a simple spreadsheet or database. Each entry must have a unique Firing ID (e.g., 2024-09-15-Cone6-Sculpture) and fields for both descriptive and prescriptive data points from your facts.
Step 2: Log Relentlessly & Specifically. After each firing, before unloading, record the descriptive reality. Who loaded it (Loader)? What were the Atmosphere Observations? Did you change clay bodies? Note the Kiln Used and any Kiln Sitter/Controller Notes. This builds your unique dataset.
Step 3: Analyze with AI for Patterns. Feed this structured historical data into an AI tool designed for analysis, like a custom GPT or an AI-powered spreadsheet. Don't ask, "Why did my glaze crawl?" Instead, query: "Cross-reference all firings where glaze crawling occurred with the recorded clay body type and bisque porosity notes." The AI will surface correlations invisible in a notebook, like crawling only happening with a specific dusty bisque.
The key takeaway is that consistency is a data problem. By systematically logging the split reality of your firing process, you transform intangible craft wisdom into a searchable, analyzable asset. You move from chasing symptoms to controlling variables, one tracked firing at a time.
Top comments (0)