1 min lesson
System design: an AI workflow under real constraints
Rebuild the parts of "System design: an AI workflow under real constraints", then say why each one matters.
Step 1 of 2
System design: an AI workflow under real constraints~60 min
Expect to design a production AI workflow inside enterprise reality. The differentiator is that you raise the unglamorous constraints yourself instead of waiting to be asked.
Constraints to surface unprompted
- Trust & auth
- SSOSingle Sign-On. One company login (usually via SAML or OIDC) instead of a separate password per tool. Press Enter for the full definition./SAMLSecurity Assertion Markup Language. The XML-era enterprise standard that powers single sign-on: your identity provider vouches for you to each app. Older than OIDC but still what many enterprise tools speak. Press Enter for the full definition., trust boundaries, what the agent is allowed to touch.
- Observability
- Tracing, metrics, alerting - how you'd see what the system did.
- Evals
- How you'd prove it works: golden tasks, pass/fail gates, regression detection.
- Failure & rollback
- Failure modes, guardrails and a clean way to roll back.
- Latency & cost
- Caching, batching, model selection, token budgeting tradeoffs.
Always answer ‘how do you know it works?’
For a code-modifying AI workflow, ‘it works’ means a concrete eval story: a set of golden tasks, a pass/fail gate and regression detection - plus the metric you'd watch in production. If you can't say how you'd measure it, you haven't designed it.