v0.52.1
July 18, 2026
Data security and a more reliable engine: items are readable only when their flow is visible to your role, sensitive fields are masked in exports and AI channels too, non-blocking AI tasks no longer lose their verdict, and published Process Maps become immutable. Plus a round of polish on the Automation Plan and AI Assistants.
-
Security
Item visibility with no shortcuts
Reading a single item's detail — including its AI overview and execution logs — now requires the flow to be visible to your role, as was already the case for lists, history and tasks. Knowing an item's direct address is no longer enough to see its data.
-
Security
Sensitive fields masked wherever they leave the app
Fields declared sensitive on a flow (e.g. tax code, diagnosis) are now redacted in the items CSV export, in MCP tool responses and for the Read-only role. People working the items still see them in the detail view, where they are needed to operate: protection kicks in on the channels that get downloaded and shared.
-
Improvement
AI task verdicts are no longer lost
When an automatic state's rules depend on the outcome of a non-blocking task, the engine now re-evaluates the rules once the task completes, instead of immediately routing the item down the default path with data not yet available. At publish time, a new warning flags rules that read the output of a non-blocking task in the same state. And if an assistant points to a deleted or deactivated AI provider, the task fails with a clear message instead of silently using a different provider than the one you chose.
-
Improvement
Published Process Maps immutable and What-if instantly meaningful
Editing the design of a published map is now blocked with a dedicated message: you create a new version, and the published one stays intact for history, comparisons and economic baselines. When entering What-if, automatable nodes come preselected: the potential saving shows up immediately, with no need to click node by node.
-
Improvement
A more solid Automation Plan
Backlog initiatives can now be deleted; creation from What-if is double-click and concurrent-submit proof; economic amounts only accept sensible values, keeping ROI and portfolio totals trustworthy. In addition, the Saving Challenge export requires a consolidated period on the server side too, item notes show a properly translated label and reject non-text values, and AI Assistants clearly report duplicate names and invalid providers.