Honest readiness: why 'coming soon' builds more trust than 'fake active'
AI platforms depend on many services. Showing configured/unconfigured/degraded honestly prevents incidents, builds trust, and helps operators sleep.
I once joined a team whose dashboard showed every model as 'active' — green dots everywhere. It looked great until a customer tried Gemini and got a cryptic 500. The API key had expired two weeks earlier, but the readiness check was hardcoded to return true. The customer lost two days debugging. That experience shaped how I think about readiness.
Four states, no more
- Configured: Dependency connected, tested, responding within acceptable latency. Green means real.
- Unconfigured: Setup required. Feature exists in code but has no production credentials. Show a setup guide, not an error.
- Degraded: Responding but with elevated latency or error rate. Works but should be used carefully. Show warning with current metrics.
- Coming soon: Not implemented. No production API promised. Show estimated timeline if available, never a fake active state.
// Cached readiness checks with timeouts
const store = new Map<string, ReadinessStatus>();
async function checkReadiness(provider: string) {
const cache = store.get(provider);
if (cache && Date.now() - cache.checkedAt < 30_000) return cache;
try {
const start = performance.now();
const res = await fetch(healthUrl(provider), { signal: AbortSignal.timeout(5000) });
const status = { state: res.ok ? "configured" : "degraded",
latencyMs: Math.round(performance.now() - start), checkedAt: Date.now() };
store.set(provider, status);
return status;
} catch {
return { state: "unconfigured", checkedAt: Date.now(), error: "Health check failed" };
}
}Updated:
Ready to ship your AI product?
Start free, route across providers, and see honest cost + readiness from day one.
Related reading
- Product
VeloxAI: the multi-model control plane for product teams
Why product teams need one API for models, agents, RAG, billing, analytics, and readiness instead of another thin provider proxy.
- Models
How to choose the right AI model for every product workflow
A battle-tested model selection framework covering cost, latency, context window, tool calling, vision, and reasoning — with real numbers and a decision matrix.
- Knowledge Base
Building a production RAG system that doesn't lie to users
A production-grade RAG pipeline needs ingestion state, chunk metadata, vector isolation, citations, queue-based indexing, and honest failure modes.