Release notes

Nightly Builds

What landed in each nightly build, before it reaches a production release. For testers, partners and anyone who wants an early look.

Subscribe to nightly notesProduction releases
Pre-release. Nightly builds run in our nightly environment for testing and early feedback. Changes here can be revised or held back, and nothing on this page is a commitment for the next production release. For what is generally available, see production releases.
AI generated. These notes are written by AI from the build's changes, and AI sometimes gets things wrong. If you have questions, please contact us through Fuuz customer support.

All nightly builds

4 Builds, Newest First

Monday, November 2, 2026 2026-11-02 nightly3 changes

JSONata editors are faster and more reliable, and detail screens with JSON fields no longer warn about unsaved changes when nothing was edited.

  • ImprovedJSONata IntelliSense now runs in the background, so the page stays responsive even with very large payloads or context. Each editor now keeps its own completions, so switching between editors (for example, between two Data Flow transform nodes) no longer replaces one editor's suggestions with another's. $$ is suggested as soon as you press Ctrl+Space. The documentation panel appears for the highlighted completion. A "Building IntelliSense…" badge shows while suggestions are loading. A "Simplified IntelliSense" badge names any bindings whose sample data was too large to analyze in full.
  • FixedDetail screens with a JSON field, such as Device, Device Gateway, Device Driver and Connector, no longer show the "unsaved changes" prompt when you leave without editing anything. The prompt still appears after you edit the JSON field.
  • FixedIn JSONata editors, a Context that is not an object (for example, "hello") no longer adds unwanted $0, $1, … IntelliSense bindings.

Wednesday, October 28, 2026 2026-10-28 nightly1 change

One fix shipped for Web flows in the Data Flow Designer.

  • FixedWhen you run a Web flow from the Data Flow Designer and it calls another Web flow with $executeFlow, that call now runs in the browser flow engine. Before, the designer sent the call to the server. Nested Web flows now behave the same in the designer as they do at runtime.

Tuesday, October 27, 2026 2026-10-27 nightly3 changes

Today's changes improve Flow Designer performance, make data model deployment safer, and add a way to open a component directly in the App Designer.

  • ImprovedYou can now link straight to a component in the App Designer. Add ?open=<Type>:<id> to the App Designer URL (for example, ?open=DataFlow:myDataFlowId) and that component opens in the workspace. If the value is unknown or malformed, the designer opens on its normal start page.
  • FixedThe Data Flow Designer no longer sends a stream of repeated requests from Data Changes nodes while you edit a flow. Flows with many Data Changes nodes should now stay responsive and put much less load on the application service. Log node level indicators still update as before.
  • FixedA data model can no longer be deployed if its id field has no unique setting. Deployment now rejects the model with a validation error. Before this fix, models like this could get through and break the application build.

Monday, October 26, 2026 2026-10-26 nightly3 changes

This release adds schema support to data flows, gives new screen elements a default data path, and fixes a Data Change Capture bug that affected scaled environments.

  • ImprovedYou can now define an input schema (JSON Schema) on a data flow's Request node and an output schema on its Response node. The flow's inputSchema and outputSchema are available through the API, and payloads are checked against them. If a payload doesn't match the schema, the flow raises an error instead of running with bad data. Existing flows that have Request/Response nodes default to an open {} schema, so they keep working as before.
  • ImprovedIn the screen designer, new form inputs and table columns now get a default dataPath based on the element's System Name. Before, a screen would error at runtime if you dropped one of these elements in and didn't set a data path.
  • FixedFixed a Data Change Capture bug in environments running more than one Data Change Capture instance. When a data model was deployed, only one instance picked up the new configuration. Because of this, change events for newly deployed models could be skipped and never recorded. Fields marked Sensitive could also be stored unredacted, and excluded-field, retention, and label-field settings were applied inconsistently. Every instance now reloads model configuration after each deployment.