Edge intelligence · nothing leaves the plant

An industrial AI agent that runs on site

A self-contained appliance running a 27-billion-parameter language model and the OpenClaw agent runtime locally. Four concurrent sessions, 40+ tokens per second in total, about 58 W average — no cloud account, no data leaving the network.

Industrial AI agent appliance, desktop enclosure with DC input and dual Ethernet ports
Reference unit: desktop enclosure, 12 V DC input, two Ethernet ports — one gigabit for the plant network, one 100 Mb for an instrument segment or an isolated cell.
27BLanguage model resident on the appliance
4Concurrent agent sessions
40+Tokens per second, aggregate
58 WAverage system power, measured

Why on site instead of in a cloud

Plant data — P&ID sets, maintenance logs, meter readings, operator notes, photographs of a failed bearing — is exactly the data that a security review will not let out of the building. An agent that needs the internet to answer a question about a pump is an agent that will not be deployed.

This appliance answers locally. The model, the documents, the conversation history and the tool results all stay inside the plant network. There is no per-token bill and no dependency on a vendor staying in business or keeping an API online.

Measured, not estimated. The performance figures on this page were recorded on our reference unit: 4-way concurrent generation at 8K context per session, 10.4–10.9 tokens/s per stream, NPU utilisation reaching 99%, hottest point 52 °C, about 1.5 GB of the 15.6 GB memory in use.

What the agent actually does

  • Answers from your documents. Manuals, commissioning reports, contracts, test sheets — indexed locally, answered with the source cited.
  • Reads what is on site. Photographs and scanned nameplates are read by on-board NPU OCR, so a picture of a pump nameplate becomes structured data.
  • Runs scheduled work. Daily energy summaries, shift handover notes, meter comparisons, overdue-maintenance lists — generated and pushed to the people who need them.
  • Talks to the systems you already have. Modbus, MQTT and HTTP interfaces for meters, PLCs and monitoring platforms.
  • Connects to your messaging. WeChat, Feishu, Telegram or email — the same agent, reachable from a phone on the plant network.

Configuration

Reference hardware — Rockchip standard family

HostRK3588, 8-core
Accelerators4 × RK1828
AI throughput20 TOPS INT8 / 40 TOPS INT4 per card
Memory15.6 GB visible, 4 × 5 GB stacked
Network1 × GbE + 1 × 100M, optional 4G/5G
Power12–36 V DC, ~58 W average
EnclosureDesktop or DIN-rail, fanless option

Software layer

  • OpenClaw agent runtime, running offline
  • Local model serving, 27B quantised to W4A16
  • Local document index with citations
  • NPU OCR for photographs and scans
  • Scheduled tasks, audit log, role-based access

Measured performance

Reference unit, four sessions at once

27B model, W4A16 quantisation, 8K context per session, four sessions generating simultaneously.

MetricMeasuredNote
Time to first token0.9–2.6 sPer session, four running in parallel
Decode speed per session10.4–10.9 tokens/sStable across the run
Aggregate throughput40+ tokens/sFour sessions combined
NPU utilisation83 → 99%Saturates as all four streams fill
Memory in use1.2–1.6 GB of 15.6 GBLeaves headroom for a larger context
Temperature32–52 °CHost 32–41 °C, accelerators 38–52 °C
Long-context mode32K per session, 2 sessionsAbout 6.5 GB per card at 32K

A single session runs faster than 10 tokens/s; the figure above is the per-stream rate with four sessions sharing the hardware. Context length trades against concurrency — the appliance is configured for whichever your workload needs.

Industrial use

Where an on-site agent earns its place

◍

Equipment condition

Vibration and acoustic data ingested locally, trend summaries and anomaly notes written in plain language for the maintenance team.

▤

Energy accounting

Meter data collected, specific energy calculated per unit of output, and the weekly saving report drafted before the meeting.

≋

Gas and ventilation

Drainage and ventilation readings summarised against limits, with exceptions flagged instead of buried in a spreadsheet.

⌁

Maintenance knowledge

Manuals, drawings and past work orders searchable in natural language — including photographs of nameplates and damage.

◷

Shift reporting

Handover notes, production summaries and incident drafts generated from the data the plant already records.

✦

Field engineering

Selection checks, pump sizing and quotation drafts produced on site, without a connection back to head office.

Deployment

The appliance is delivered configured, with the model loaded and the agent runtime installed. Commissioning is a network drop and an account setup, not a software project.

  • Runs fully offline; internet is optional, for updates and remote support only
  • Local accounts and role-based access, with an audit log of every agent action
  • Document ingestion done on site — files are not uploaded anywhere
  • Optional site-to-site VPN for vendor support, enabled by the plant when needed
  • Model and knowledge base can be updated from a USB device, no network required

What it is not

It is not a replacement for a plant historian, a DCS or a safety system. It reads, summarises and drafts; it does not close a control loop, and it is not connected to protection logic.

It is also not an unlimited model. A 27B model on edge silicon is a capable assistant for engineering and reporting work — it is not a frontier model, and we do not present it as one. The value is that it runs here, on your data, at a fixed cost, with no network dependency.

Standard hardware, engineered deployment

The appliance is built on Rockchip's standard edge product family — an RK3588 host with RK1828 accelerator modules, the same silicon any integrator can buy. The hardware is not the product.

What we deliver is everything above it: the model chosen and quantised for the duty, the OpenClaw agent runtime, the document index, the tool connections, the prompts and the guardrails that make it answer like a plant engineer instead of a chatbot — plus the people who pick up the phone when it does not.

Same box, your brand. For distributors and system integrators we ship the appliance configured and tested, with your nameplate, your start screen and your support contact. You sell it into your market; we keep the runtime and the knowledge base current.

Where we are taking it

Our first export markets are the ASEAN countries and Southeast Asia, where a plant often has the machines and the meters but not the bandwidth, the budget or the data-protection clearance for a cloud AI service. An appliance that runs on 58 W, needs no internet and answers in the local language fits those sites.

  • English and Chinese interfaces today; further languages on request
  • Operates on a plant LAN with no external connection
  • Supplied configured — commissioning is a network drop, not an integration project
  • Distributor and OEM terms available for the region

Bring a laptop to the plant, leave with a working agent.

Tell us the documents you want it to read and the questions you want answered. We will demonstrate it on your own files, offline.

Arrange a demonstration