Incident process
When something goes wrong in your workspace — a service crash, a degraded response, a security event — Han AI follows a documented incident process. This page summarises what you can expect.
What counts as an incident
| Severity | Examples |
|---|---|
| Major | Telegram bot offline; dashboard unreachable; runtime crash loop; data loss; suspected Personal Data Breach. |
| Minor | A single tool degraded; a scheduled job missed; transient errors that recover automatically. |
How you are notified
- Operational incidents that affect your workspace are surfaced to you in Telegram by your Han AI bot, and on the dashboard.
- Personal Data Breach notification follows DPA §6 — without undue delay and in any event within seventy-two hours of Han AI becoming aware. The notification describes the nature of the breach, the categories and approximate number of records, the likely consequences, and the mitigating measures.
Response targets
TODO: confirm formal response-time targets and SLA tier table.
What we do
- Acknowledge and confirm the incident is real.
- Stabilise — restore service or contain the issue.
- Investigate root cause.
- Communicate with you on findings, impact, and remediation.
- Post-incident: document the cause and the durable fix.
What you can do
- For an active incident, contact your engagement owner directly. See support.
- For a suspected security issue, mark the report urgent so it routes to the on-call channel.