The scale is the source of truth
Readings are captured from the device and stored as received. A posted ticket always points back to the raw reading it came from, so a dispute about a load is settled by the record rather than by memory.
← Rivl AI / Proof The plant that knows
The plant that knows
what is in the silo.
Rivl Plant Manager is operations software for a bulk materials facility. It reads the truck scale, keeps inventory by silo and quality class, raises an alert when a number moves out of range, and answers plant questions from the plant's own records through Ask Rivl.
First deployed at a materials facility in Monahans, Texas.
The sign in screen as it runs today at the first plant. The scale house state, the loads released and the tons shipped are live before anyone logs in.
What it does One record of what came in,
One record of what came in,
what went out and what is left.
Nothing here is exotic. What is unusual is that the scale, the silos, the orders and the alerts share one record, and that a plant question can be answered from that record rather than from the person who happens to remember.
Scale integration
Tickets read from the truck scale and posted to inventory without retyping, with the raw reading kept beside the posted value.
Inventory by silo and class
Each silo carries its own balance by quality class, so the office knows what can be sold before the next truck is loaded.
Orders and loading
Customer orders tied to outbound tickets, so a load that leaves the yard is already matched to who ordered it and what it weighed.
Alerts
A reading out of range, a silo below its reorder level or an order short of material raises a notice to the role that owns it, not to a named person.
Ask Rivl
Plain language questions answered from the plant's own tickets and inventory, with the records the answer came from shown beside it.
Roles, not names
Permissions, notifications and escalation are defined by role. People are assigned to roles in settings, because people change and roles do not.
Under the hood Built for a yard,
Built for a yard,
not a demonstration.
A plant does not stop for software. The system has to take a bad scale reading, a truck that arrives before its order, and an operator on a phone in the dust, and still keep one honest number.
Inventory cannot go negative by accident
Balances are enforced in the database, not only in the application, so a bug or a direct write cannot produce a silo balance the yard knows is impossible.
Answers grounded in the plant records
Ask Rivl answers from tickets and inventory rows and shows them. When the records do not cover the question, it says so rather than filling the gap.
Notices with a delivery record
Every alert and email is stored with the outcome the provider actually returned, so the system reports what happened rather than what it attempted.
5,000Tickets a year, projected from the first plant's recorded throughput
300 STPer silo at the first deployment; silos, grades and customers are configuration, never code
15 hoursA plant day, 7:00 AM to 10:00 PM, set per plant and read by every alert and digest
1 teamThe same people who scope a client engagement built and operate this
What it proves Every operation has a scale,
Every operation has a scale,
a bin and a question nobody can answer fast.
Underneath the plant vocabulary this is a device integration, an inventory ledger with enforced rules, and an assistant grounded in operating records. That combination shows up in warehouses, laboratories, clinics and depots.
Device and system integration
Instruments, meters and legacy systems with no interface, captured once and kept beside the posted record.
Balances that cannot drift
Any count that has to be right: stock, cash, units, beds, seats. Enforced at the data layer so an audit finding is a query rather than an investigation.
Retrieval over your own operations
Your records made answerable in the words your staff use, with the source shown so an answer can be traced.
Escalation by role
Notices routed to the role that owns the exception, so a change of staff never breaks the escalation path.
Rivl AI
Your operation has a number nobody trusts.
Thirty minutes. Tell us what gets measured, who writes it down and where it goes wrong, and we will tell you whether this pattern fits or whether the constraint is somewhere else entirely.