judged.systems
The judgment API platform for support systems.
36 followers
The judgment API platform for support systems.
36 followers
Support systems call judged.systems when a ticket needs a judgment. The helpdesk stays the system of record. You send the ticket. A pack defines the questions: queue, urgency, refund request, policy risk. Back comes a choice and its probabilities, a score on a rubric, or the probability a statement is true. Your thresholds return accept or review. A failed call stays in review. The ticket is redacted first. The evaluation is stored as it happened. A label on a miss leaves it untouched.
Free
Launch Team
Viktor.comAn AI coworker that actually does the work
Promoted
Hey, Iām Oleksandr š . I built judged.systems.
A ticket comes back as a choice and its probabilities, a score on a rubric, or the probability a statement is true. Your thresholds turn that into accept or review. A failed call stays in review.
A published pack is frozen. A label on a miss does not rewrite the evaluation.
Where would you refuse to let a rule auto-accept?
answering your own question: anywhere the ticket author has an incentive to word things so the classifier lets them through. refund requests and policy risk are exactly that, the person writing the ticket is sometimes trying to get classified as low-risk on purpose, so the probability score you'd threshold on is also the thing being gamed. queue routing or urgency, nobody's lying to you about that, it's a fine place to auto-accept. the moment the input is adversarial rather than just noisy, a high-confidence score stops being reassuring