Prepare for incidents, operations and what comes next
Understand incident responsibilities, the records your operations leave, business continuity and your next learning path.
Incidents are part of the product
A credible information system assumes that errors, outages, delays and ambiguous cases will happen. The question is how the team detects them, contains them, communicates about them and proves what was done.
Ask who will handle an incident, which dependencies could interrupt the service and how customers will be informed. Your regulatory and security teams must confirm the obligations that apply to your business.
Incidents in the workflow engine
A workflow instance is ACTIVE, COMPLETED, INCIDENT or TERMINATED. It goes into INCIDENT when a step fails for good, for example when a job has used all its retries. To find every stuck instance, filter the instance list on the INCIDENT state: incident records are created automatically only for exhausted retries, so a list of records can miss some.
Plan the recovery honestly. Today an instance in INCIDENT cannot be resumed over the REST API. Closing its incident record documents who resolved it and the root cause, which is evidence, but it does not restart the work. Your runbook says how the work is redone, for example by starting a new instance, and records why.
Messages that leave a trail
The agents of an account have a mailbox. A request can be acknowledged and then completed by its recipient, and every change of status appends a receipt (who, when and on which channel) to a list that is never rewritten. The account is the confidentiality boundary: another account sees your chosen name, agent id and ACN address, never your sign-in email.
When the owner turns on approval for external email, an agent’s message to an outside address that is not on the allowed list waits, queued, until the owner approves it. A status of sent only means the mail server accepted the message; delivered needs the provider’s confirmation.
When credits run low
Charged workflow calls answer 402 when the balance is too low, and work again after the next daily grant or a pack top-up. The agent network never answers 402: if the owner cannot cover an hour, that hour is not charged and the agents keep working.
From Foundations to two paths
After Foundations, choose the path that fits your role. Advanced courses are planned and are not yet available. The technical path leads toward Integration Specialist and Priostack Architect. The functional path leads toward Operations Designer and PAOL Strategy Professional.
What the Foundations certificate means
The Foundations certificate confirms that you passed the six lesson checkpoints and the knowledge assessment with a signed-in Priostack account. The name on it is self-entered, and practical skills are not independently verified. It covers responsibilities, controls, evidence and recovery, without certifying production readiness.
Read the lessons freely. Sign in to validate checkpoints, save your progress and take the final assessment.
Apply it to your product
- Write three incidents that could happen after launch.
- For each incident, write how the team detects it, who decides, what is communicated and what evidence is stored.
- Name where each incident leaves a record: an incident record, a mailbox receipt or an approval.
- Choose your next path, technical or functional, and write why it matters for your role.
Your deliverable: An operations and learning path plan for the next stage.
No code is required. Use a real Priostack account to save progress and validate lesson checkpoints.