How we work

From a first call to the whole fleet, one step at a time.

You start small, on the data you already have, and you decide at every step. Below: what we deliver, and what we need from you.

  1. 1

    We talk about your asset

    A 30-minute call on the machine, its failures and the systems you run.

    You bring

    • Asset type and manufacturer
    • The control systems and historians you have
    • Roughly how many similar units

    You get

    • An honest view on whether predictive maintenance pays off on that asset
    • The connector level for each of your systems

  2. 2

    Demo and data check

    Sibilla on your asset type, on simulated data; then, under NDA, a first look at a sample of your data.

    You bring

    • Read access or an export of a few months of the relevant tags
    • The tag list, as it is

    You get

    • A data quality report: missing, frozen, duplicated and out-of-range points
    • Your tags mapped to the failure modes they can reveal

  3. 3

    Pilot on a few units

    The Blueprint of your asset, on your data, on-premise or in the cloud. Goals and success criteria are written down before we start.

    You bring

    • Read-only access to the historian or controllers
    • A contact in maintenance for the reviews
    • A virtual machine with Docker, if on-premise

    You get

    • Sibilla configured, KDIs running, Teneuro where data is missing
    • Periodic reviews of what Sibilla sees, with your team

  4. 4

    Scale to the fleet

    The same Blueprint cloned across similar units; your team trained to run it.

    You bring

    • The list of units and sites
    • The people who will use it

    You get

    • Fleet view and Asset Health for every unit
    • Integration with your CMMS or ERP, if you want it Project-based
Book the first call Is predictive worth it on your asset?

For IT and OT

The questions your IT and OT colleagues will ask.

Where does our data stay?

Where you decide: in your network, in your cloud (Azure, AWS, Google Cloud or any Docker host) or in ours. It is the same platform in all three cases.

Does Sibilla write to our control systems?

No. Sibilla reads, read-only, from your historian, SCADA or controllers and sends no commands to the plant. No operational data leaves your network when it is installed on-premise.

Does it need an internet connection?

No. Installed on-premise, Sibilla runs inside your network without internet access; weather data can be switched off.

What do you need from IT?

A read-only account on the historian or the OPC UA server, or an export of the relevant tags. For an on-premise installation, a virtual machine with Docker; we send the requirements before the pilot.

How does Teneuro send its data?

Over MQTT, to a broker that can sit in your plant or in the cloud. There is no PC at the machine.

Who can see what?

Viewer, Editor and Superuser roles per organisation, inherited along the tree of your sites. Single sign-on with your corporate identity provider is Project-based .

What do AI agents do with our plant?

Agents work through Sibilla's API and MCP server: they read, configure and run, always with a preview, your explicit confirmation and an audit log of every action. Detection stays deterministic and verifiable.

Who owns the data and the models?

It is agreed in writing before the pilot, under NDA. We use your data only for the work you ask us to do.

Can it open work orders in SAP or Maximo?

Yes, as an integration built on your system and processes: Project-based . Our team has designed maintenance processes in SAP on a global scale.

SAP, Maximo, Azure, AWS, Google Cloud and the other names cited are trademarks of their respective owners; they are cited only to describe compatibility.

Start with one asset.

Tell us the asset type, manufacturer and control systems: we'll prepare a demo for your case.

Request a demo on your asset