Don't take my word for it
Playground
Small backend patterns, running live against this site's own API — the same building blocks behind rate-limited public APIs and streaming LLM products. Poke them.
Rate limiter, live
A sliding-window limiter guards this endpoint: 5 requests per 10 seconds per visitor. Burst past it and watch the 429s.
5 per 10s
No requests yet.
Token streaming, live
One HTTP response, flushed chunk by chunk — the mechanics behind every fast-feeling LLM interface.
Output appears here, token by token.
Your latency to this site
Six real round-trips from your browser to this site's API, measured client-side. The first one pays the connection cost.
Want the full version of any of these in your product? Let's talk.