Application Design

New

1. Containers can scroll when their content overflows. A new Scroll on Overflow option makes a container scroll vertically instead of growing past the edge of the screen. Use it to keep buttons or headers permanently visible alongside a long form.
   

Improved

1. Query extra fields without writing GraphQL. Additional Query Fields is now simply Query Fields, and it uses a field picker instead of free text. Data Path is now Data Model Field, and property descriptions were rewritten to explain what each one actually does.
2. New screen elements no longer query fields you didn't ask for. Form inputs and table columns now start with no query fields selected. Previously a new element fell back to its data path, which could produce an error if that field didn't exist on the model. Screens generated from Screen Designer templates are unchanged for now.
3. A cleaner Data Model editor. Dropdowns replace comboboxes, switches replace checkboxes, system fields are collapsed by default, and Retention Days is hidden unless Data Change Capture is on. Model names are now checked in the form itself, so an invalid name is caught as you type rather than rejected when you save.
   

Fixed

1. GraphQL and Script editors keep your work when you switch tabs. In Application Designer, leaving a GraphQL or Script Editor tab and coming back reset it to the default query and discarded your changes. Both editors now restore exactly what you left. Closing a tab and opening a new one still starts fresh.
2. Edit Properties opens the element you clicked. In the Screen Designer Structure tab, Edit Properties previously opened the properties of whichever element was already selected rather than the one whose menu you used. It now works as expected.

Orchestration

New

1. Run and breakpoint controls on the node itself. Debug and source nodes now always show an action bar on the canvas, so you can run a node or toggle its breakpoint without opening a menu. Nodes also lift to the front when you hover or click them, so they no longer hide behind their neighbours.
   

Improved

1. Auto Layout produces flows you don't have to clean up. Data Flow Auto Layout was rewritten from scratch. It now arranges flows cleanly left to right and stops inserting link points that then had to be deleted by hand. Give it a try!
2. Faster package linting. The package linter now reuses model dependency information instead of recalculating it on every run.
   

Fixed

1. Opening an editable flow after a view-only flow. Going from a read-only or substituted flow into an editable one could throw an error, and the flow wouldn't execute until you touched a node. Moving a node also failed to mark the flow as changed, leaving Save disabled.
   
2. Node configuration screens corrected across the designer. Several data flow nodes had errors in the definitions that drove their configurations, which could make those nodes difficult or impossible for AI tooling to use while building flows. Running flows were never affected.
   

System Administration

New

1. (Beta) Control which AI tools each user can call. MCP is now a selectable base in the access control policy editor, with an Execute Tool action and every available tool listed individually. Administrators can grant a user all tools, a specific subset, or none. Users without an MCP policy cannot call tools.
   

Improved

1. One dialog for switching tenant, role, and environment. The footer's Tenant, Role, and Environment buttons now open the same Switch Tenant or Role dialog the drawer menu uses, so switching works identically everywhere. Role switching no longer sends you to a separate page, and Switch Role has been removed from the avatar menu.
   
2. Records can no longer be created with a blank ID. A create request or a trigger that produced an empty ID previously saved a record with a blank ID. A generated ID is now used instead. Existing records with blank IDs continue to work as they did before.
   

Fixed

1. API Access users are kept out of the web app. An API Access user could sign in and find an empty app with no accessible screens and a Sign Out button that misbehaved. API Access and Gateway Access generic user accounts now stay on the login page with a message explaining that only API access is authorized.
   
2. Enterprise sync completes the Application Definitions step. This step of the enterprise sync failed on an email validation error, and blank records left behind by it caused errors on the Applications screen. Both are resolved and the empty records removed.
   

User Experience

Improved

1. Search finds what you meant to type. The header search bar now uses fuzzy matching rather than requiring an exact substring, so app designer finds Application Designer and word order no longer matters - making it easier for users to find what they're looking for if they don't know the exact name (or just misspell it).
   
2. Bold text finally looks right. Some weights of bold text was being computed client-side by the browser rather than loaded. The full range now loads, making certain weights of bold headings and labels noticeably sharper.
   

Fixed

1. Navigation menu keeps its open groups. Pinning or unpinning the left navigation drawer collapsed every expanded group. Expanded groups now survive pinning, going fullscreen, switching navigators, and resizing the window. They still reset when you change tenant or role.
   

Integration & Edge Gateway

New

1. Ignition connectivity. Fuuz now connects to Inductive Automation's Ignition platform through a new Edge Gateway driver and Ignition module. The Ignition module, installed on an Ignition gateway, links it to a Fuuz Device Gateway; the Ignition tag provider then appears in Fuuz as a device, so you can browse and subscribe to tags, read and write values, and call device functions exactly as you would with any other device. The module also adds functions to Ignition tag scripts to run Fuuz data flows, queries, and mutations within Ignition applications.
   
2. QuestDB time-series driver. A new time-series device type lets the Device Gateway write measurements to, and read them back from, your own QuestDB instance at the edge. Writes are acknowledged and retried, so readings survive an unreliable network.
   

Fixed

1. Clear message when a gateway API key lacks permission. Signing in to a Device Gateway with an API key that didn't have the required permissions previously produced an endless stream of errors. It now shows a single clear message. A genuine connection problem still retries, and signs you in once the connection returns.