This is the first week covered by these updates, so it’s a short one — it picks up from 17 July.
New features
Automations can now run on a schedule. A new Schedule trigger joins the existing trigger list, so an automation can fire on a recurring timetable tied to a batch or a script rather than waiting for a call or a webhook. You give it a five-field schedule expression, a timezone, the batch or script it acts on, and limits for how long a run may take and how late it may start. Schedules must run at least sixty minutes apart. See Automation triggers.
A schedule that switches itself off tells you why, with one-click re-enable. If the batch behind a scheduled automation is no longer eligible, the script goes inactive, or a run fails or times out, the schedule disables itself and a banner explains the reason — for example “A scheduled run failed or timed out.” A Re-enable schedule button gets it going again once you’ve fixed the cause. See Execution log.
Automation pipelines can run a Code step. A new Code transform sits alongside the LLM, field mapping and template options, letting you write sandboxed Python to reshape data part way through a pipeline — reading and writing mapped fields and variables, upserting contacts and objects, routing contacts onto lists, and observing or prioritising a batch. See Code step and Automation steps.
Connect external AI applications to your Voxworks scripts. A new Settings → MCP Clients page lists every application you’ve authorised to read and revise your scripts and flows, with one-click revoke. Connecting an app takes you through a Voxworks authorisation screen where you choose the team it may act in and see exactly what it will be allowed to do — for example “Read your scripts and flows” and “Create new script revisions (every change is compiled and versioned)”.
Improvements
LLM automation steps gained a fuller set of options. You can now set the Reasoning effort (Low, Medium, High), write a System Prompt, define a typed Output Schema and map the result into variables with a Response Map. See Automation steps.
The automation execution log shows far more about each run. New columns cover the trigger time, when the run was scheduled for, attempts, the reason it fired, latency, the step path, outputs, fields written and more, and runs can now show as Running or Cancelled. See Execution log.
