It cannot see
Dozens of protocols, thousands of tags, and three systems that disagree about the same event. Every integration is a one-off.
See the Context EngineYour factory, in a language AI understands.
MaestroHub is the industrial context platform: it connects every machine and business system in your plant and turns their data into one governed language, so your people and AI can see the plant, understand it and act on it, inside your rules and on your own infrastructure.
120+ machines and systems
AI agents
Claude, or any MCP clientMaestro agentsoonApps and analytics
Dashboards and appsData platforms and BI- 120+ machines and systemsConnect what the plant already runs, from PLCs and historians to SAP.
- Small footprintOne binary on one edge box is enough to start. No cluster required.
- Ready for AI agentsYour agents build pipelines and dashboards in MaestroHub over MCP.
- One codebaseThe same product from a single edge device to Kubernetes.
Why industrial AI stalls
AI is smart enough to understand your factory.It just has no context.
You bought the AI. It still cannot tell you what a stop cost, because all it sees is numbers: nothing gave your data one meaning, one map, or permission to act.
It cannot understand
A value has no neighbours. Nothing says what feeds what, what controls what, or what stops if this stops.
See the Knowledge GraphIt may not act
Nobody lets an agent touch a line without knowing who it acts for, what it may do, and what it did.
See agents and governanceThe architecture
One layer between your machines and everything that runs on them.
- 1A machine sendsOne of 120+ machines and systems
- 2Context Engine works on itThree things, on every value
- 3People and AI readA value they can trust
ns=2;s=P101.VIB3.4No name. No unit. No quality.MaestroHubContext Engine
Names itPump 1, vibration
Adds the unit and checks qualitymm/s, good
Records where it came fromOPC UA, PLC-2
Pump 1, vibrationGood
3.4 mm/s
from PLC-2
and 80+ more connectors, reaching 120+ kinds of machines and systems.
One namespace for every machine and system.
More than 90 connectors reach over 120 kinds of machines and systems. Context Engine gives every value its schema, its quality and where it came from. Take a destination offline and watch the buffer drain.
- 1The plant is mappedThings, and the typed links between them
- 2You askIn plain words, or an agent asks
- 3It answers with the pathRows you can check, and how it got there
Line 2, demo plant. Illustrative.
What stops if Pump 1 stops?
Also ask
What does Tank 1 feed?
What changed on Line 2 since Tuesday?
Which pumps vibrate above 3 mm/s?
2 things stop. 1 can take over.
StopsFiller 2fed by Pump 1
StopsCapper 2fed by Filler 2
StandbyPump 2also feeds Filler 2
Walked 2 hops downstream over "feeds". As of now.
The plant as things and links.
Knowledge Graph holds every thing in the plant and how it relates: what feeds what, what controls what, what stops if this stops, and what the plant looked like on any past day. Illustrative answer on a demo plant.
- 1Someone asksAn agent, or an app a domain expert built
- 2MaestroHub checks and asks firstIt acts only as a named person
- 3It acts, and records itEvery step lands in the audit trail
Claude, through MCP
Capper 2's gearbox is running hot. Schedule the service for Thursday and raise the work order.
Or ask through
The Maestro agent insidesoon
An App Studio appsoon
Acting forRyu Kaplan, maintenance lead
Read the historianAllowed
Write a work orderNeeds approval
RKRyu approves. Scoped grant, two hours.
Work order createdCapper 2 gearbox service, Thursday 14:00
Audit trail
06:02agent proposesrecorded
06:04Ryu Kaplan approvesrecorded
06:04ran as Ryu Kaplanrecorded
Chain verified
Agents and apps that ask first.
Bring your own MCP client today. Coming soon: the Maestro agent inside the product, and App Studio, where a domain expert describes the application they need and it is built on the whole platform. It all runs inside that person's permissions, and every step is audited.
The product
Three products, one platform.
Deployment
Governed by architecture, and by policy.
It is your plant. MaestroHub runs where your data is, your tools stay where they are, and nothing leaves your network unless you send it.
Your network
MaestroHub runs hereOne binary at the edge, or Kubernetes at site level- Your plant dataEvery value and its lineage
- Credentials and keysNever sent to us
- The audit trailKept and verified by you
- History windowYour system of record stays yours
MaestroHub, the companyPublishes releases with SBOMs you can check without us, and gives support on your terms.Trusted by manufacturers.
European home appliance leader
Planning, PLC and quality data in one hub, with live KPI dashboards and alerts before issues escalate.
15%operational efficiency
10%output increase
Automotive leader
Edge data and operator feedback feed anomaly models that warn supervisors before a CNC tool fails.
95%prediction accuracy
$22Ksaved per CNC a year
Chemical manufacturing leader
Distillation SCADA and ERP data joined to show steam and energy cost per batch in real time.
15%better steam use
8 tsteam saved a day
Reference architecture
The whole platform, on one page.
What comes in from the plant, what MaestroHub does with it, and who uses it. Follow the numbers.
Your plantAny vendor, any age
PLCs and controllers
CNC machines and robots
SCADA and historians
MES and ERP, such as SAP
Databases and files
Cloud and IoT services
MaestroHubOne platform, running inside your network13 releases, Feb to Aug 202637 engineering rules, each from a real defect
2Context EngineSee the plantLive2.6.5, Aug 2026
Connect
- 90+ connectors
- 120+ machines and systems
- 22 of them write back, under audit
Give it context
- One namespace for the plant
- Schemas and units
- Quality and provenance on every value
Move it
- Pipelines and functions
- Store and forward
- Buffers on disk when a link drops, then replays in order
Serve it
- MQTT and REST
- Dashboards and alerts
- MCP server for agents
3Knowledge GraphUnderstand itComing soonin development and testing
- Every thing in the plant and how it relates
- Ask in plain words, get the path back
- The plant as it was on any past day
4Agents and App StudioAct on itLiveagent accessComing soonApp Studio
- Your own MCP client, or the Maestro agent (coming soon)
- Apps a domain expert describes, App Studio builds
- Every action inside one person's permissions
5Governed write-backActions return to the plant as a named person, checked first and recorded. A data path, not a control loop.
Governance on every path
- Role-based access
- Single sign-on: OIDC, SAML, LDAP
- Audit trail you can verify
Fleet Manager
- A registry of every MaestroHub instance
- Licences, health, identity settings
- Edge sites connect outbound only
One codebase, five ways to run
- A single binary at the edge
- A container, or a Siemens Industrial Edge app
- Kubernetes, or a fleet hub with edge sites
Who uses itPeople, agents and systems
Your peopleOperatorsEngineersMaintenance
AI agentsClaude, or any MCP clientThe Maestro agent, coming soon
Apps and analyticsDashboards and BIApp Studio apps, coming soon
Data platformsSnowflake, DatabricksKafka, your historian
- One trusted picture of the plant
- AI that asks before it acts
- Runs on your own infrastructure
- One product, edge to Kubernetes
Compliance and control
Ready for the plant, and for IT.
- ISO 27001 certifiedOur information security management is certified to ISO/IEC 27001.
- Access you can reason aboutRole-based access, ownership, sharing and time-bound grants, for people and agents alike. Single sign-on over OIDC, SAML and LDAP.
- An audit trail you can verifyHash-chained and tamper-evident, with its boundary stated first, and one call that checks the chain.
- A supply chain you can inspectCosign-signed images and SBOM attestations on every release. Verify them without us.
Free trial
Try it yourself, today.
Download MaestroHub and run it on your own machine. You get every feature, Enterprise ones included, in two-hour sessions that you can restart as often as you like, with no licence and no sales call.
Windows, macOS and Linux
- 1DownloadGet MaestroHub from the portal and start it. No licence key.
- 2Connect somethingYour own PLC or broker, or the free Digital Factory Simulator: a four-station line with 100+ tags.
- 3See it workBuild a pipeline and a dashboard on live data, then decide whether the six-week pilot is worth it.
Questions, answered plainly.
What is MaestroHub?
MaestroHub is an industrial data platform. It connects machines and business systems, gives every value one meaning in a governed namespace, holds the plant as a knowledge graph, and lets AI agents and applications act on the factory inside permissions and an audit trail. It has three products: Context Engine, which is live, and Knowledge Graph and App Studio, which are coming soon.
What is available today, and what is still being built?
Context Engine, the MCP server for agents, access control and the audit trail are live in release 2.6.5 (August 2026). Release 3.0 follows in October 2026, with the approval gate. Knowledge Graph, App Studio and the Maestro agent are not part of 3.0: they are in development and testing, and coming soon.
How do AI agents connect to MaestroHub?
Two ways. You can point your own MCP client, such as Claude or Cursor, at the MaestroHub MCP server. Coming soon, you will also be able to use the Maestro agent inside the product. In both cases the agent sees only what the person it acts for is allowed to see, and every action is recorded.
Does MaestroHub replace my SCADA, historian or MES?
No. MaestroHub sits above the systems you already own. Ignition, your historian, your PLCs and your MES stay where they are. MaestroHub reads from them, writes back under permissions and audit, and keeps an operational window of history of up to about seven days by licence.
Where does MaestroHub run?
On your own infrastructure. The same codebase runs as a single binary on a laptop or at the edge, as a container, as a Siemens Industrial Edge app, or on Kubernetes at site level. Your data does not leave your network unless you send it.
How is MaestroHub priced?
Per instance, not per tag. There are Foundation and Enterprise editions. The usual way to start is a six-week pilot on one virtual machine with no production writes.
Build on MaestroHub.
One VM, no production writes, six weeks. If it does not earn its keep, you lost a VM.
Behind this text: three demo plants as one knowledge graph. 287 things, 511 links.