MayaBehavior
“Maya: I paste the request into our support sheet, then copy it into the roadmap doc if three other people ask.”
A product bet, under pressure
Find the signal.
Keep the quote.
Jev makes small, typed choices. Zentrik’s code maps them back to the exact words that shaped the call.
Evidence map
Illustrative walkthrough · no provider call
The product bet
Product teams need one shared inbox for customer feedback.
The call, under pressure
6 of 6 turns connect to this bet. Read both sides before you decide.
SSupports the bet
3MayaFriction
“Maya: Finding the original conversation. It can be in email, Slack, or call notes.”
Inspect this judgment
MayaFriction
“Maya: Support can see tickets, but they cannot see which product bet we are testing.”
Inspect this judgment
QAdds a condition
2MayaConstraint
“Maya: The same request can mean different things. We should not merge them just because the wording matches.”
Inspect this judgment
MayaConstraint
“Maya: If auto-grouping hides the difference between a workaround and a request, I stop trusting it.”
Inspect this judgment
↗Pushes back
1MayaPreference
“Maya: Do not make another inbox. Put the source next to the decision, and let me check the quote.”
Inspect this judgment
·Adds context
0Nothing landed here. That is worth noticing.
This walkthrough is illustrative. The shown labels are not a Jev prediction; no transcript leaves this page.
01 · Keep the source
The model returns choices, not rewritten testimony. Zentrik attaches each label to the verbatim turn.
02 · Look for the antiverse
A good bet map makes caveats and pushback visible, not just the quotes that make a feature sound right.
03 · People decide
These are prompts for review, not proof of demand, a product-wide finding, or an automatic roadmap choice.