Trust
Grounded, auditable, assurance-grade. If we cannot prove what the system did and why, we do not ship it.
About Slicium
The hardest factory floor deserves the best autonomy.
Protein processing feeds the planet on one of the largest, most labour-intensive and most dangerous factory floors in existence. We think that work should be done by machines that can see.
Every animal is anatomically unique. That is why automation has failed here.
Fixed machines repeat a motion. A centimetre off leaves saleable meat on the bone or over-trims premium product. The only way to automate this work is to perceive each carcass and adapt to it — which is exactly what became possible in the last few years, and exactly what Slicium is built to do.
Principles
What we hold to.
Autonomy
We perform the work rather than assist with it. A copilot that watches the line does not solve a labour crisis.
Outcomes
We price and measure on value delivered — yield, giveaway, labour, safety — not on seats or dashboards.
Vertical depth
Built for protein specifically. Species anatomy, cut specs and food-safety standards are the product, not a configuration file.
The opportunity
The market we are building into.
$25B
Total addressable market across processing automation, inspection and plant software
$6B
Serviceable market for autonomous cutting, grading, detection and twins
$330M
Three-year obtainable market
15%
Annual category growth amid an acute labour crisis
Why now
Three curves crossed at once.
Perception finally became good enough
Fusing RGB, depth, hyperspectral and X-ray at line speed with GPU-accelerated inference makes per-carcass understanding practical. Five years ago the compute, the sensors and the models were not there together.
- Edge GPU inference at 30–120 FPS
- Hyperspectral sensing at industrial cost
- Simulation-generated rare-event training data
- Robotics rated for washdown environments
The labour crisis stopped being cyclical
Turnover, repetitive-strain and laceration injuries and chronic understaffing on the cut floor are structural. Processors are not choosing between people and autonomy — they are choosing between autonomy and lost throughput.
- Structural cut-floor understaffing
- Injury and turnover costs rising
- Razor-thin margins on every gram
- Food-safety and recall risk increasing
Roadmap
Where we are going.
MVP with design partners
One workflow agent on one line, plant-edge deployed, with shadow-mode baselining.
Advisory mode GA
Advisory operation across cutting, grading and detection with operator correction capture.
Graduated autonomy + twin
Supervised and graduated autonomy gated on twin validation — hit the yield before the cut.
Second module + enterprise
Multi-agent plant coverage, fleet management and enterprise readiness across sites.
“The craft of a master butcher took thirty years to build and disappears when they retire. Encoding it is the most valuable thing this industry can do with AI.”
How we work
A company shaped by the floor.
Engineers in washdown gear
Our team commissions on site, at 4am, in the cold and the wet. Nobody at Slicium designs a perception system for a plant they have never stood in. The product is better because the feedback loop is short and physical.
Evidence over enthusiasm
We would rather lose a pilot than defend a number the customer’s own systems do not support.
Customer IP is sacred
Cut programs and specs are the customer’s competitive advantage. They never leak across tenants.
Independent
Founder-led, pre-IPO, building original technology. Not a consultancy, reseller or division. [PLACEHOLDER]
Long-term vision
Become the autonomous operations layer for the world’s protein processors — every carcass cut, graded, screened and packed on Slicium.
Working with
Northfold PoultryMeridian PorkCascadia ProteinBaltic Fillet Co.Ardenne FoodsHarbour & Sons Seafood
Design-partner and pilot plants. Names shown are representative programme cohorts. [PLACEHOLDER]
Responsibility
Autonomy in food comes with obligations.
SOC 2 Type II (in progress)HACCP-alignedUSDA-FSISEU 853/2004GDPRISO 27001 (planned)
Per-tenant isolation
Every plant gets isolated data, models and vector stores. Cut recipes and species models never cross a tenant boundary — federated learning shares patterns, never your product IP.
Assurance-grade audit trail
Immutable logs link sensor frames, model version, autonomy level, approvals, overrides and outcomes for every agent action — built for recall defence and model governance.
Fail-safe by design
Blade, robot and line stops are deterministic and independent of the cloud. Graceful degradation returns the cell to a safe state if perception confidence drops.
Newsroom
Recent from Slicium.
Why we built the twin before the product
Simulation was not a nice-to-have. Without it, no processor would let an agent near a blade.
The real cost of weight giveaway
A gram per pack does not sound like much until you multiply it by a year of production.
We are hiring physical-AI engineers
Perception, robotics, simulation and plant-edge infrastructure roles are open now.
Come build the autonomous protein plant.
Whether you run a plant, build the equipment, or write the perception code — we want to talk.