Dynamic at the speed of static with instant warmup after every deploy

Spectare has two caches, checked in turn, and both are held for seven days. On a warm request the visitor's classification is served from a Redis cache keyed by the org and signal summary, and the assembled page from a separate Redis cache keyed by intent type, audience, and stage. When both hit, the full server-rendered response completes in around 200 milliseconds with no AI calls: the personalised page is inlined into the initial HTML, before any JavaScript runs.

On a cold request the server classifies the visitor with Claude and assembles the page, then writes both caches so the next visitor with the same signal pattern skips the AI entirely. Because the common intent patterns and each page's canonical entry are pre-warmed ahead of traffic, even a first-time visitor usually lands on a warm result. Cold assembly is the exception, not the rule, and it is self-healing: any pattern that runs once is warm for everyone after.