PLATFORM · OPINION · Updated 2026-09-10 · Createrun Engineering

On-prem in 2026: choosing where your software runs

The deployment decision starts with the application, the people operating it and the connections it needs. Visual development and AI assistance do not remove those responsibilities.

Start with the boundary

Createrun installs the application runtime and process engine on servers you operate. Activation and package downloads connect to Createrun. A customer-controlled instance can help an organization manage infrastructure and operational data, while its team owns patching, backups, monitoring and recovery.

There are two separate uses of that instance. An application built on the Createrun framework uses its runtime. Development tools running on the instance can also help organize work on a project in another stack. The location of those tools does not determine where the resulting software must run.

When to examine an on-prem deployment

A restricted network, an existing data estate or a contractual hosting requirement may make local infrastructure a practical choice. Evaluate the actual requirement before selecting a deployment model. A regulation's name or a server's country alone does not establish that a deployment meets every obligation.

For supplier information, review the agreed access and processing terms. Moving data to a hosted service can require a contract review, but it does not invariably mean rewriting every NDA. Operating locally also leaves responsibilities for access, retention and any data sent through integrations.

Connectivity is a separate question

List which operations need an external connection: activation, package acquisition, identity services, business integrations and AI calls. Test the required offline workflow against the installer and package version you intend to use. Installing an application on premises does not automatically give a mobile user offline functionality.

An instance on your own servers can still send selected context to an external AI provider. Choosing your own API key is not the same as running the model on your own infrastructure. Check provider settings and the information included in each call.

What to include in the evaluation

  • Install a specified version with documented prerequisites and record the steps.
  • Verify the package and its dependencies before use.
  • Test identity and permissions with the roles your team will use.
  • Apply an update in a test environment and rehearse recovery for data and external effects.
  • Identify who handles backups, incident response and ongoing support.

The security overview describes platform controls and their stated status. It is a starting point for an architecture review, rather than a substitute for testing your configuration.

Where the framework helps

Using existing forms, data access and workflow infrastructure can reduce the amount of custom infrastructure an application needs. Developers can use the Editor directly, extend with code or use supported AI operations. The benefit depends on how well that framework fits the application.

We do not present a measured months-to-weeks improvement or a token-saving percentage here. A useful pilot compares accepted functionality, failed attempts, human effort and operating cost for the same scope.

Make the operating choice explicit

Compare a customer-operated instance with a managed service using the responsibilities your team can actually support. Include implementation, hosting and AI usage in the budget, and agree the supported deployment and license terms. Packaging can support distribution between environments; environment configuration and data movement still need planning.

Book a demo and describe your deployment requirements. We can scope a working evaluation around the application and network boundary you need.

Back to blog