Jev AI: How do you make small decisions inside an AI workflow?

[bellwood]

A support ticket arrives. Your workflow needs to decide which team should handle it, how urgent it is, and whether a person should review it before anything happens.

I’m building Jev AI for decisions like these. You send the current state and define the questions you need answered. Jev returns structured choices or scores, along with probability and confidence signals. Your application then decides what action to take.

For example, one request could ask:
• Which team should receive this ticket?
• How urgent is it?
• Does this need human review?

I’m particularly interested in how developers handle low-confidence results in real workflows. Would you set a fixed threshold, combine the result with business rules, or send every uncertain case to a person?

I’d welcome critical feedback, especially on where this kind of decision model would or wouldn’t fit your workflow.