Tasks
Tasks run on a schedule and notify you when something is worth your attention. You pick the context to run on, write instructions for what matters, and set how often to run. When a task finds something, it reaches out with a Notification.
What are Tasks?
Tasks run the same kind of check you'd ask for in a chat, but on repeat. They execute in the background on your schedule and reach out with a Notification when they find something worth your attention.
Each task has:
- Context: the insight, dashboard, connection, database, segment, or semantic model the task operates on
- A type: the kind of work the task does, such as anomaly detection, forecasting, or a general task driven by your instructions
- Instructions: plain-language guidance that tells the task what to focus on (for most task types)
- A schedule: when the task should run
- A status: whether it is currently active or paused
Task Types
Altertable supports several kinds of tasks, each tuned for a different kind of analysis.
| Type | What it does |
|---|---|
| Anomaly detection | Spots unusual patterns on an insight on your schedule and notifies you |
| Forecast | Generates forecasts for an insight on your schedule and notifies you |
| General task | Runs your instructions against the context on your schedule and notifies you on anything noteworthy |
The type you can pick depends on the context you attach the task to. When you open an insight, for example, you'll see the task types that work with insights (anomaly detection, forecast, general task). A general task works across insights, dashboards, connections, databases, segments, and semantic models, and runs the same way for each: the task reviews the context against your instructions and reaches out when it finds something noteworthy.
Instructions
Instructions tell the task what to focus on. Be specific about thresholds, patterns, and what deserves a notification. Clear instructions produce more useful output and keep notifications from becoming noisy.
You can update the instructions on a task at any time from its detail page. Changes take effect on the next run. If you're not sure where to start, write something simple and refine it after reviewing the first few runs. You can also trigger a run manually at any time to test your changes immediately.
Scheduling
Tasks run on a schedule defined by a cron expression. You can change the schedule from the task's detail page.
In the app, you can select a preset that gets converted to a cron expression:
| Interval | When it runs |
|---|---|
| Hourly | At a chosen minute past every hour |
| Daily | At a chosen hour each day (UTC) |
| Weekly | On a chosen day and hour each week (UTC) |
| Custom | Any cron expression you define |
You can also run any active task manually at any time from the task menu, outside its schedule.
Status
A task is either Active or Paused. You can toggle the status from the task's detail page or from the task menu.
| Status | Behavior |
|---|---|
| Active | Runs on its schedule |
| Paused | Stops running until you turn it back on. Configuration and history stay intact |
Pausing is the right choice when you want to hold a task temporarily without losing its configuration or history.
Run History
Every execution of a task is recorded as a run. Open the Run history tab on a task to review what ran and when.
| Status | Meaning |
|---|---|
| Queued | Accepted by the scheduler, waiting to execute |
| Running | Currently executing |
| Succeeded | Completed without error. May have created notifications |
| Failed | Stopped because of an error |
| Cancelled | Was stopped or replaced by a newer run |
Each run records the instructions that were active at the time it executed. If you have edited the instructions since a run, the run detail will show a warning so you can compare what ran versus what you have now.
Suggestions
As you open an entity, Altertable proposes task suggestions based on the entity and the other tasks. Suggestions appear in the task panel with a pre-filled title, instructions, and schedule.
Selecting a suggestion opens the task creation form pre-populated so you can review and adjust everything before saving. If none of the suggestions fit, dismiss them and start from scratch.
Suggestions are available for insights, dashboards, connections, databases, segments, and semantic models.
Creating a Task
- Open an insight, dashboard, connection, database, segment, or semantic model you want the task to run on.
- Start a new task or select one of the suggestions.
- Write clear instructions. The quality of the results depends on this: include thresholds, what to watch for, what counts as noteworthy, and any context that helps narrow the focus. The more specific you are, the more relevant notifications you'll get.
- Set a schedule that matches how often your data changes.
- Save the task. It will run on its schedule and appear in your task list.
A task title is generated from the task configuration. You can rename it at any time from the task detail page.
Learn More
- Agents: the automation workers that run your tasks
- Notifications: the findings that tasks produce
- Memories: knowledge agents carry across runs
- Dashboards: attach tasks to monitor KPIs
- Insights: the insights and analyses tasks run on