Every firing, perfectly made.
Porcelio runs the parts of a ceramics plant that still depend on somebody watching. Body preparation, pressing, drying, glazing and kiln firing are perceived and controlled by one system, which brings down defects, downgrades, energy use and manual sorting.
Tile, sanitaryware and technical ceramics run on kilns, and kilns run on judgement. A plant leans on a handful of people who can read a firing curve and adjust before a batch goes off tone. That knowledge is real, it is scarce, and it retires.
The timing is what moved us. Firing is one of the most energy-intensive steps in building materials, so carbon pricing now lands on every plant directly. Large-format slabs and digital decoration have raised the cost of a defect. Meanwhile high-resolution tone and planarity vision, kiln instrumentation and rugged edge inference have only recently become good enough to close the loop rather than just report on it.
Porcelio does not try to replace the plant. It watches pressing, drying, glazing and firing, then closes the loop on the decisions that were previously made by feel. The result shows up where a plant manager already looks: fewer downgrades, less breakage, and a lower energy bill per square metre.
What convinced us was the wedge. Defect and tone vision is narrow enough to install in one line and prove in a quarter. Everything else the company wants to do sits on top of that first camera.
Each agent earns its place on its own. Together they are the reason the firing curve stops drifting.
Milling and slip preparation held to particle size, moisture and rheology, so the body arriving at the press is the same body it was last week.
Force, thickness, density and green strength controlled at the press, where most downstream defects are actually created.
Drying rate and glaze application tuned to prevent the cracks, pinholes and crawling that only reveal themselves after the burn.
Firing curves that adjust to humidity, body moisture and load rather than holding a setting written years ago.
Defect, tone, caliber and planarity inspection at production speed, so a bad batch is caught at the press rather than at the sorting table.
Robotic glazing, kiln loading, tone and caliber sorting, and packing. The repetitive, hot and unpleasant jobs, done without a rota.
The part that is hard to copy is not any single model. It is the loop, and the firing data that accumulates every time it runs.
The firing curve, shrinkage and glaze behaviour are simulated before the burn, so the target caliber and tone are hit on the first pass rather than sorted out afterwards.
Most incumbents ship a scanner that assists one task and reports. Porcelio writes back to plant controls and robotic cells under human-supervised autonomy.
Every firing teaches the system something a point tool never sees, because no point tool spans body, press, dryer, glaze line and kiln at once.
Co-founder and CEO
Twelve years in ceramics manufacturing, most of it on the floor of tile and sanitaryware plants in Staffordshire and northern Italy. He can tell you what a kiln does at three in the morning when the humidity moves, which is the problem the company was built around.
Co-founder and CTO
Vision and control systems engineer. Shipped edge inference for industrial robotics before this, and has the useful habit of testing a model against a running production line rather than a held-out set.
Co-founder and COO
Spent nine years running plant rollouts for industrial automation vendors, which means he has installed hardware during a shutdown window and knows how little patience a production line has for a pilot. He owns deployment, commissioning and the customer relationship after the first camera goes up.
Company profile prepared by 99sa Ventures. Team details and role descriptions are indicative and are updated as the company confirms them publicly.
Applied AI in an industry that still runs on judgement and spreadsheets is close to the centre of what we look for.