What univedo does for a production site
univedo is not another application. It is the layer underneath: the shared picture of the production and the rules by which applications may use it. This page explains the platform in plain terms. The technical details live in the documentation.
Cloud, on premise or private cloud, and free to switch
- Cloud. Nothing to install. Applications and data are hosted and operated for you.
- On premise. A guided setup turns standard servers into a univedo installation. No configuration files, single sign-on, only an outbound internet connection needed. Updates come from the store; certificates, nodes and connectivity are monitored. Data never leaves the site.
- Private cloud. The same installation in your own cloud tenant or data centre, under your security policies and your keys.
- Switch when needs change. All three run the same software, so moving between them, from a cloud pilot to your own servers or the other way round, is a planned migration rather than a new project. Data, applications and approvals move together, and users keep working as before.
- Mix per site. A pilot in the cloud, one plant on premise, another in your private cloud, all on the same platform and updated from the same store.
- AI under your control. Every AI call from any application goes through one component that you configure: which providers, which models, which monthly limits. Run models on site when data must not leave the plant.
- Remote support. Support sessions are tunnelled through the platform, with your consent, without opening inbound ports.
One shared picture
Orders, parts, programs, tools, machines, locations, people: on univedo they exist once, and every application sees the same thing. An ERP link contributes orders, a tool management application contributes tools, a machine connector contributes what happens at the machine. No copies, no nightly synchronisation jobs, no arguing about which system is right.
The picture is allowed to grow. When a vendor extends it, the change is recorded as a version, and applications built against the previous picture keep working. That is what makes it possible for many vendors to share one platform without breaking each other.
Processes that cross systems
Real processes do not respect system boundaries. A production order needs a program from CAM, tools from the tool room, a slot on a machine, and a report back to planning. On univedo, applications hand these steps to each other as events. The machine connector learns that a program is ready, the tool application learns that a machine needs a tool, planning learns that a lot is finished.
The handover is reliable. If an application is down, it receives what it missed when it comes back. Every event says which application caused it, so the trail from order to finished part is complete.
New applications join safely
Whether an application is bought from a vendor, built by an integrator, or prompted into existence, the same rule applies: it starts in its own space, it is checked against the shared picture, an expert approves it, and only then may it change production data. The platform verifies the approval itself, on every write.
Applications ship through the Univedo Store. Customers see for each one that it has been approved and what part of the shared picture it works with. Vendors get a single target to build for instead of one integration per customer.
The safe path for prompted applicationsThe Univedo Store
The store is where applications are published, licensed, bought and updated. A vendor offers bundles of functions; a customer picks the bundle that fits, per machine and per user, in the cloud or as an on-premise licence. The installation on site fetches what it needs from the store and keeps it current.
Visit the storeFor the technically minded
How the shared picture is modelled, how versions and approvals work, which client libraries exist and how to build on the platform is documented in uspec, the part of univedo where the picture is governed. Start with the technical introduction.
See univedo with your own production data
In a one-hour session we connect a sample of your data, walk through the shared picture, and build a first prompted application in front of you.