Buzz vs BAND
Bench log · how to set up buzz app, buzz.xyz setup

Bench: time to first agent reply, Buzz

Last verified: 2026-08-19

Desk draft · not yet bench-measured

The method below is fixed and published. Timers read Not yet measured until the run completes.

This entry measures one number: the elapsed time from clicking download on buzz.xyz to reading the first agent response inside a channel on a Block-hosted relay. It replaces the desk estimate behind the Buzz time-to-real-work score. Desk-drafted from the bench brief. Method is fixed and published below. Numbers land here after the run, and until then every timer on this page reads Not yet measured.

Time to first agent reply
Not yet measured

Fresh macOS 15 machine, Block-hosted relay, method fixed 2026-08-19

What is the method?

One fresh machine, one account, one stopwatch. No prior Buzz install, no restored identity, no pre-warmed cache.

  • Machine: fresh macOS 15 install, no prior Buzz artifacts
  • Tier: free early access, Block-hosted relay
  • Start: click on the download control at buzz.xyz
  • Stop: first agent message rendered in a channel we created
  • Splits recorded at: download complete, install complete, account created, keypair stored, community created, agent added

Where did the minutes go?

Not yet measured. The split table publishes with the run, and the largest split gets its own paragraph, because that is the number an operator plans around.

What did Developer Preview status cost us?

Not yet measured. Preview products fail in ways that are not the product's steady state, so we log each stumble, note whether we retried, and report both a clean-path time and a real-world time.

Does this move a score?

Possibly. The Buzz time-to-real-work cell is a desk estimate with a HIGH confidence flag on the shape of the flow, not on the duration. A measured run replaces the estimate and any score change ships the same day with a changelog note.

Related

Method fixed 2026-08-19 · DESK REVIEW · PROVISIONAL