Scheduled task automation — cron jobs run by AI

Frequently asked questions

• What is scheduled task automation at Nordix Systems?+

  Scheduled task automation at Nordix Systems means a named AI scheduler agent inside Nordix BIOS owns every recurring job in your business — inventory reorders, newsletter sends, accounting exports, schedule generation, daily report posts. Operators describe each job in plain language and the scheduler runs it on time, monitors the result, retries on transient failure, and reports outcomes in the same chat thread used for the rest of operations.

• How is this different from a cron job?+

  Cron executes a fixed shell command at a fixed time. The Nordix scheduler agent executes intent — the underlying steps are AI-planned and adapt to context. If a step fails (supplier down, API rate-limited, missing image), the agent improvises within policy, escalates if needed, and writes a human-readable explanation. Cron has no notion of intent; the Nordix scheduler does.

• Why not use Zapier or Make for these workflows?+

  Zapier and Make connect fixed triggers to fixed actions and break when a system changes its shape. The Nordix scheduler agent reasons about the goal of the job, picks the right system today, and re-plans if a dependency moves. Operators do not maintain visual workflows — they edit a one-paragraph job description in chat. Failure rates on long-running jobs drop sharply as a result.

• How do I see and edit my scheduled jobs?+

  Type ‘list my scheduled jobs’ in chat. Nordix BIOS replies with a categorized list — daily, weekly, monthly, ad hoc — each with the last run status. To edit, reply ‘change the newsletter job to ship on Wednesdays at 09:00 instead of Tuesdays at 08:00’. The scheduler updates the job, confirms the new cadence, and runs a dry-run preview before the first live execution at the new time.

• What happens when a scheduled job fails?+

  When a job fails, the scheduler posts a structured incident to chat: the job name, the failure reason, the systems it touched, the customer impact if any, and one or two suggested fixes. Operators reply ‘retry’, ‘skip’, or with a corrective instruction. Recurring failures trigger a deeper review where the scheduler proposes a permanent policy change for operator approval.

• Is there an audit trail of every scheduled run?+

  Yes. Every run is logged with the job version, the trigger time, the actual start and end times, the operator who created or last edited the job, every system call the agent made, and the human-readable outcome. The log retains 13 months by default and exports to a customer-managed data warehouse on a configurable schedule for compliance use.