Design a system where several specialised agents collaborate to respond to a production incident: one triages the alert, one investigates logs and metrics, one correlates recent deploys and config changes, one drafts the remediation, and one keeps the incident channel updated.
They work in parallel under time pressure, on a system that is already broken. The remediation agent can make things considerably worse.
Build the architecture on a canvas: place the components, configure them, connect them into a data flow, and write a short reason for each one. The AI reviewer grades your design against a rubric written specifically for this problem.
Two investigating agents reach contradictory conclusions. Describe exactly what happens next in your design.
Which actions may the system take without a human, and how is that boundary defined and enforced?
The incident is in the observability stack itself, so the agents' tools return partial or wrong data. How does the system behave?
What does the handoff to a human contain when the system cannot diagnose the incident?
Minimum 9 components · needs a wide desktop screen