Platform / Ship and run
What is deployed where, what is running, what failed, and the log line that says why.
Product screens recreated with synthetic data.
Scripts run on managed compute where the plan includes it. Team watches pipelines. It does not include this fleet.
▸ status requested → launching task invoice-extract started mounted dataset ap-inbox read-only read INV-20841.pdf extracted vendor=Acme amount=12480 amount > 10000 · paused for review waiting on Review · not posted
DevOps is where an app goes live and keeps running. A release is an approval record — version, knowledge version, gates — executed against UAT first, then production, with a roll back on the same screen. Workloads you start from the console move through requested, launching, running, done, error, or stopped. Live logs are on the card. Stop and delete are on the card. You should not need a second cloud console to know whether last night’s job is still alive.
Feature health is the extra question. A process can be up while the thing it is supposed to do is failing. The fleet list is allowed to say both. Incidents you want a person to hear about belong on Reachout once that module is live; the history of runs belongs in Athena.
DevOps does not approve a workflow write and it does not define an agent. A green workload is not an audit trail. The audit trail for a commit is the agentOS record. This screen is health and logs.
Whoever starts scripts and containers on the platform and has to answer, the same day, whether they finished.
Watching pipeline work is part of Platform. Running your own scripts and the heavier compute classes start on Platform Business, where compute is turned on. Team does not include a compute fleet.
Pricing →Synthetic example. Not a customer.
A container step has been “running” longer than the documents justify. You open the workload, read the log, and stop it. The pipeline’s gate is a separate object — stopping the container does not approve the invoices that were waiting. It stops the compute. The documents still need a clean run.
Requested through stopped, on the workload, not in a ticket.
The terminal output is one click from the fleet list.
Stop and delete from the same screen that showed you the problem.
Services and workloads with a live status.
When launching turns into error, the output is on the card.
Feature health is there so a live process can still be a failed feature.
Stop or delete from the console. Then rerun the workflow when the cause is fixed.
Platform Start is the self-serve canvas. A first pipeline we stand up with you is delivery, priced apart from the subscription.