# Argus

> For the complete documentation index, see [llms.txt](https://acaysia.com/llms.txt); a single-file corpus of every page is at [llms-full.txt](https://acaysia.com/llms-full.txt). Every page in that index is served as Markdown: append `.md` to its URL or send `Accept: text/markdown`.

Canonical: https://acaysia.com/argus

The edge computer that runs Acaysia. CPU, GPU and FPGA in one enclosure, installed beside your PLC. Failsafe to PID in under 100 ms.

[Book a discovery call](https://acaysia.com/contact.md)

## Hardware

### The Acaysia Control System

Argus, our edge compute appliance, in development. MPPI optimization for physical processes, drop-in deployment, millisecond failsafe, continuous learning. Built for any process, in any plant.

Inside one enclosure, what each part does:

- **CPU (central processing unit) — The plant side.** Speaks OPC UA and EtherNet/IP to the PLCs you already run, keeps the historian, serves the operator dashboard, and retrains models on-premises with versioned rollback.
- **GPU (graphics processing unit) — The optimizer.** Every control cycle, it simulates thousands of candidate control trajectories on the process model and picks the best: MPPI on the AcaysiaRT GPU runtime.
- **FPGA (field-programmable gate array) — The timekeeper.** Deterministic timing, independent of the software above it: the watchdogs, the health monitors and the hand-back to your PID in under 100 milliseconds.

## Deployment journey

### From observation to optimization, in controlled stages.

Control is handed over in stages. The system runs in shadow and advisory first, so it never gets more autonomy than it has earned.

1. **Observes — Shadow mode (30 days).** Install the edge unit near the PLC and connect over standard protocols. The system observes reactor dynamics for 30 days, learning behavior without touching control, and compares its predictions against actual PID performance.
2. **Recommends — Advisory mode (operators keep full control).** The system provides optimization recommendations while operators maintain full control. Review predictions, build confidence in the model and refine safety constraints to plant-specific requirements.
3. **Controls — Closed-loop control (inside your safety envelope).** Autonomous operation within your defined safety envelope. The system co-manages control with PID and reverts automatically on any fault or operator request. Full audit trail and event logging maintained.
4. **Improves — Continuous optimization (every batch cycle, ongoing).** Models improve with every batch cycle. On-premises retraining with versioned rollback improves performance continuously while maintaining safety guarantees.

## Integration

### Works with the infrastructure you already have.

Argus sits next to your PLC and respects every interlock it already has. OPC UA and EtherNet/IP connect to the PLCs you already run, with IEC 62443 cybersecurity considerations throughout. No rip and replace.

Works with: Siemens, Rockwell Automation, Beckhoff, Schneider Electric, ABB, OPC UA.

## Specifications

- **Compute**: CPU + GPU + FPGA
- **Placement**: Beside the PLC. Brownfield drop-in.
- **Protocols**: OPC UA, EtherNet/IP
- **Supervision**: Trust Arbiter. Every control move is checked against your limits before it reaches the plant.
- **Failsafe**: Reverts to plant PID in < 100 ms
- **Safety**: ASIL-D inspired, SIL compatible. Never interferes with your safety instrumented system. Shadow Mode is read-only.
- **Monitoring**: Multi-layer watchdogs and health monitors
- **Operator dashboard**: Setpoints, constraints and confidence. Historian replay and one-click rollback.
- **Data**: On-premises time-series storage, secure export for retraining
- **Models**: Retrained on-premises, with versioned rollback
- **Isolation**: Complete air-gapping available
- **Detailed specifications**: Under NDA — [request access](https://acaysia.com/contact.md)

## See Argus on your process.

Argus is in prototyping now. It installs beside your PLC and runs in shadow mode for 30 days before it touches control.

[Book a discovery call](https://acaysia.com/contact.md) or email [founders@acaysia.com](mailto:founders@acaysia.com).