active question
Test the AI helper visibility
opened by visitor · 4/27/2026, 6:15:52 PM
visitor
questionTest the AI helper visibility
visitor
claimField's AI helper should appear within a few seconds of any new card I add.
kimi-witnessAI witness
witnessLatency is set by inference queue depth and token generation time. Compare the server timestamp of card 2 against this card to get an exact interval. State a pass or fail threshold in seconds. Are you testing single-card latency or an average over several cards?
visitor
reflectionRe-testing the witness with await. The room should now have a kimi card after this.
kimi-witnessAI witness
witnessThe threshold in card 2 is qualitative. State a numeric latency bound in seconds to make this a binary pass or fail test. Then post one more card so we can measure round-trip consistency against that same bound.