Monitor one operational area without opening a long-running chat.
Automations are powered by KAIROS internally, but the product surface is named Automations. Select a store, choose an inspection template and review the resulting operational history.
Automations run focused background inspections and record results that can become scoped engineering work; they do not silently deploy storefront changes.
Use Automations for focused background inspection.
An Automation continues independently of an open agent conversation and records a run. It is suitable for repeatable checks where the expected output is a report or finding set.
Choose theme health, performance, accessibility or conversion friction.
Each template explains the category, inspection scope and expected output. Select the smallest template that matches the operational question you need answered.
Current inspection templates are read-only by design.
The run analyzes and reports. It does not deploy storefront changes. Turn a suitable result into a normal scoped session when implementation is required.
Track active, completed, failed and interrupted runs.
The Automations page separates active work from completed results and attention states. A newly triggered run appears immediately using the run id returned by the backend.
Use the output to decide whether engineering work is justified.
A report should identify evidence and next actions. It does not guarantee a business result and should not become production code without the normal scope, staging, QA and approval workflow.
Automation runs can use Engineering Run capacity.
The current plan and Billing page show available capacity. TaskerArmy reserves sufficient capacity before starting qualifying work and releases or finalizes it according to the run outcome.
Related TaskerArmy pages
Direct answers before installation
Is KAIROS a separate product?
No. KAIROS is the internal automation engine. The customer-facing navigation and workflow are named Automations.
Can an Automation deploy a fix?
The current templates described here are inspections. Implementation follows the normal session and approval workflow.
Can I leave the page after starting a run?
Yes. The run continues in the background and its status appears in operational history.
What should I do with a failed run?
Review the attention state, confirm the store connection and retry only after understanding the failure. Use Troubleshooting or support when it persists.
Connect the store. Start with evidence.
Run the connected-store audit, review prioritized findings and decide which bounded engineering job should move to staging first.