Governed operations, plant floor to back office
Tokoaido reads your machines and your business systems, finds how the work really runs, then runs it with digital workers that stop for a person when it matters and record who decided what, and why.
Design-partner phase · runs in your cloud or on-prem
The platform
An approval is a card someone signs. An audit record says who signed and why. A boundary is a place the architecture does not go. Here is each one.
Condition · OT
The line draws over its ISO 20816-3 zones. Crossing into zone C turns the asset to Warning and puts a proposal in front of the person who signs.
Replace drive-end bearing in Saturday's 06:00 window. Reliability Sentinel, score 0.86.
Closed loop · OT → IT
Sense, diagnose, propose, approve, execute, verify. The same six steps for every closed-loop action, whether the trigger is a bearing or an invoice.
Governed execution
Risk decides whether a person must sign. The decision goes into the audit log with who signed, why, the risk and what happened next.
Earned autonomy
Agents you build or connect start advisory. Each earns scope from its measured record and is moved back when its reliability drops.
Promoted to Autonomous · 10 actions, trust 0.94
Safety boundary
Writes go through SCADA and MES with approval. A direct PLC write, or any contact with the Safety Instrumented System, is blocked in the architecture.
Neutral by design
Tokoaido keeps a working model of the records a use case needs, in your deployment or in our EU-hosted service, governs the work on top, and writes results back where they already belong.
Examples of systems that stay authoritative. Each is connected in a pilot.
Connectivity
Live today: 41 business systems over REST and OAuth, and machines over OPC UA and MQTT Sparkplug B. The other machine protocols are in the catalog on sample data and are built for your stack in a pilot. Every connector says whether it runs on sample data or live credentials.
connectors in the catalog: 49 IT with a live REST, OAuth, SQL, SFTP, S3, Azure Blob or ArcGIS path, 27 OT with live OPC UA, MTConnect and Modbus paths and sample data for the rest
agent tiers: sensing, diagnostic, optimizer, executor, orchestrator
autonomy levels: advisory, supervised, autonomous
direct agent writes to a PLC or the safety system
Compared
Typical of each category: machine-health tools see the plant and hand someone a report. Workflow suites run the office and rarely see the plant. The work in between is still done by hand.
| Machine-health tools | Workflow suites | Tokoaido | |
|---|---|---|---|
| Sees machines | yes | usually not | yes, live over OPC UA and MQTT Sparkplug B |
| Sees business systems | rarely | yes | yes |
| Carries out the action | usually raises an alert | inside its own platform | across both, with approval for anything high-risk |
| Where your data lives | often the vendor's cloud | often the vendor's platform | stays in your systems; we keep a working copy of what we read |
| How an agent gets autonomy | n/a | usually configured | earned from its record, and revoked |
| Proof of what happened | logs | logs | an audit log of who decided what, and why |
How a pilot runs
Three steps, in order. Each one ends with something you can check before you commit to the next.
Questions
Yes. It takes live machine telemetry over OPC UA and MQTT Sparkplug B, works out the as-run process, and can turn a machine condition into a governed business action, such as a rising vibration signature into a maintenance work order. MTConnect, Modbus and vendor controllers are in the catalog on sample data and are built for your stack in a pilot.
No, by construction. Writes go through SCADA or MES with human approval. A direct PLC write, or any contact with the Safety Instrumented System, is blocked in the architecture. No prompt, policy or agent can override it.
Agents you build or connect start advisory. Each one moves from advisory to supervised to autonomous on a measured track record of confirmed findings and approved, effective actions, and is moved back automatically when its reliability drops.
No. Your business and plant systems stay the system of record. Tokoaido keeps a working copy of the records a use case needs and governs the work on top, either in our EU-hosted service or deployed in your own cloud or on-prem.
We are in a design-partner phase with a working product. Production deployments are set up with design partners as connectors are turned on and verified for each stack.
Operations X-Ray
Pick a line you know well. We read its machines and the systems around it and show you where the work really goes.
Prefer to look first? Ask for a guided walkthrough, or sign in if you have an account.