1990 – today
The old baseline
SCADA and the MES. A central room watches a plant that cannot think. Intelligence lives in one place because, for thirty years, it had nowhere else to live.
Cybarete The next industrial baseline
SCADA was the baseline for the factory that could not think. BASE is the baseline for the factory that can.
The factory of the future will not be supervised from a room. It will be alive: thousands of small programs that sense, negotiate, and act on their own.
Stellenbosch, South Africa / Colorado, USA
Three factory orders at the centre, a slurry pump, a conveyor drive unit, and a robot gripper, each break down into sub-assemblies and then parts. Each part contracts production cells from one shared pool around the edge: milling, turning, a quality lab, a paint shop, welding, cutting, forming, and additive. Each cell is made of its own machines and operators. One order is highlighted at a time, showing its network of tasks and the cells it uses.
The shift
Now
The new baseline
BASE. A plant made of thousands of small programs, one per machine, per crew, per order. They negotiate work directly and keep thinking at the edge when the network drops.
The thesis
How a plant learns to run itself
The argument in five steps, grounded in the systems science of Herbert Simon: why the future of industry is negotiated, not commanded, and what it takes to build it.
- 01
The plant becomes a society
Not a control room watching dumb machines, but thousands of small programs, one per machine, per crew, per order, each deciding for itself.
- 02
It negotiates, it is not commanded
No central mind can run a live plant. Work is announced, bid for, and awarded between agents at runtime. Herbert Simon proved the command room impossible sixty years ago.
- 03
Grounded in an economy of energy
Negotiation needs a currency, and on a real plant it is energy, time, and attention. The plant does not solve for a perfect schedule. It anneals toward its promises.
- 04
The infrastructure does not exist yet
Today’s agent stacks die on a crash and cannot run forever. SCADA was built to watch a plant, not to host a society of agents. The foundation is missing.
- 05
BASE is that foundation
Built on the BEAM, the runtime made for systems that cannot go down. Stable parts that survive the interruption, coordinating into a whole.
Why BASE
An architecture, not a wrapper
BASE is not a model with tools bolted on. It is a runtime, on the only foundation built for programs that run forever and cannot fail all at once.
Control path
The control path never depends on a model
Deterministic connectors move the metal. Intelligence sits above them as a supervisor that proposes and reflects. It never holds the wheel.
No orchestrator
The runtime is the orchestrator
Every other framework bolts a graph engine or a crew manager on top. BASE deletes that layer. Hierarchy, memory, and resumption come free from the BEAM.
World model
A belief system, not a memory store
Each agent holds evidence-grounded beliefs it revises in place, cited to what it saw. Not a log of the past. A working model of its world.
Goal-integrity
Stability is architectural
Many small grounded edits anneal toward the goal, so no single step can knock the agent off it. The assurance an operator needs before a program touches a machine.
Beyond the factory
Same runtime, a different world
Swap orders for missions and machines for rovers. A colony that needs water breaks the goal down, and each step contracts the assets it needs: power, rovers, storage.
An off-world colony. The mission to bring back 100 kg of water breaks down into exploring, mining, retrieval, and storage, alongside missions to keep the habitat powered and resupply the greenhouse. Every step contracts units from one shared pool: a power station, a survey unit, a mining unit, a haul unit, and a water plant, each made of its own rovers, equipment, and crew.
In the field
Proven
BASE has been deployed in underground mines, refined through more than four M.Eng research projects, and is now the foundation of a new fleet management platform.
We are looking for case study partners. If you run a plant, a fleet, or a site that the network does not always reach, we would like to hear from you.
Become a case study → Mines deployed underground
4+ M.Eng projects
Now fleet management
Who is building it
Setting the standard now, while the window is open
Cybarete is a deep-tech company out of Stellenbosch. BASE began as a doctorate in holonic systems and has been in the field ever since. The baseline for the agentic factory will be set in the next few years, by someone. We intend it to be built to be forked and owned, not locked behind a vendor.
About Cybarete →