This post is a working demonstration of the three Living Content update mechanisms. Each block below updates through a different path. Refresh the page after running the demo scripts to see them change in place.
1. Reactive template (auto-applies)
The block below contains placeholders like {{winner}} and {{winnerPct}}%. When the poll's leader changes, the insight bridge auto-applies a new template and re-resolves placeholders. No approval queue.
AI Gateway leads at 68.4%, 42.1 points ahead of Fluid Compute (Functions) in this 19-vote snapshot of most-used Vercel features.
2. Proactive variant switch (auto-applies)
The block below has three pre-written variants tied to poll conditions. When the leader changes, the proactive evaluator switches the active variant. Authors approved the variants at creation time; no per-update approval.
AI Gateway has overtaken the rest. That signals the audience here is building AI features as their primary workload, not bolting them onto an existing app — provider failover and unified billing pay off when AI is the product, not a feature.
3. Monitored-source recommendation (requires approval)
The block below cites a fact about Vercel's default function timeout. When the monitored source's stored value changes, a stale-claim recommendation is created as pending. A human approves the new prose; only then does the block update.
Vercel functions on Hobby and Pro plans now run for up to 300 seconds by default — Fluid Compute raised the ceiling.
Run npx tsx scripts/demo-poll-vote.ts to drive blocks 1 and 2.
Run npx tsx scripts/demo-monitored-source-change.ts to drive block 3.