AI quality control · ceramic & porcelain
Inspect every tile.Not every tenth one.
tyle.tech puts computer vision and process models on your production line — catching glaze defects, tuning kiln curves, and lifting first-pass yield without slowing the belt.
Running on lines producing 1.2M m² of tile per month.
From batch to box · one model per stage
Where the model sits on your line
Tile defects compound — a moisture error at the mixer becomes a crack at the kiln. tyle.tech places a dedicated model at each stage so problems are caught where they start, not after firing.
Batch & mix
Body-composition and moisture models hold the slip within spec before a single tile is pressed.
Press & form
Dimensional vision flags warping and thickness drift the moment the die starts to wander.
Glaze
High-speed cameras catch pinholes, pooling, and color variance at full belt speed.
Fire
Kiln-curve optimization trims energy per m² and reduces thermal cracking and crazing.
Sort & box
Final grading and shade-lot matching so every pallet ships as a true single batch.
Measured across deployments
What the line looks like after
The toolkit
Six models, one line
Each capability ships as a module that drops onto existing belts and PLCs — no rebuild, no line stoppage.
Surface defect detection
Pinholes, cracks, glaze pooling, chips, and color variance flagged at belt speed with sub-millimeter resolution.
Kiln curve optimization
Models the firing profile against energy use and defect rate, then recommends curve adjustments per body and glaze.
Predictive maintenance
Reads vibration and thermal signatures off presses and rollers to schedule service before a failure stops the line.
Shade & lot sorting
Groups fired tiles into true color lots so installers never open two boxes that don't match.
Yield & batch analytics
Traces every reject back to its stage and shift, so the same defect doesn't reappear next run.
Energy & waste reduction
Connects firing, drying, and scrap data into one number: cost and carbon per finished square meter.
Line audit
Bring us one week of your reject data.
We'll map where your yield is leaking and model the lift before you commit to anything. Most audits pay for the first quarter of deployment.