LANDMARK · 6 MIN · BUILD

Fallbacks and graceful degradation

After this landmark, you can design a fallback path for an AI feature so a model outage degrades the product gracefully instead of breaking it outright.

This concept is shared, but the Everyday lens is less central here.

You can still explore it. We’re showing the shared explanation and a related practical view without hiding the knowledge.

SHARED FOUNDATION

AI model providers have outages, rate limits, and latency spikes like any external dependency, but teams that would never ship a payment feature with no fallback for a down payment processor sometimes ship an AI feature with no fallback for a down model API. A resilient design has a plan for each failure mode: a timeout with a clear message rather than an infinite spinner, a secondary model or provider to fail over to for critical paths, and a degraded-but-functional mode (cached results, a simpler rule-based response, or clearly telling the user the feature is temporarily unavailable) rather than a broken page.

This is the canonical concept. It stays the same across learner lenses so personalization never changes the underlying facts.

EVERYDAY LENS

What this looks like for you

When an AI feature you rely on suddenly errors out or hangs, that’s usually a fallback gap, not a fundamental limit; well-designed products would degrade to something useful instead of failing outright.

MAKE A DECISION

An AI-powered search feature on a retail site calls a third-party model API. That API has an outage during a high-traffic sale. What should happen?

CARRY THISFor one AI feature in a critical path you build or rely on, write down what currently happens if the model API times out. Is that the behavior you’d actually want?
CONCEPTS IN THE INDEX