Small decisions. Interesting possibilities.Submit contentSubmit

Incident triage in Box

🆕@typesafeai dropped Jev this week, a new frontier model optimized for decisions. We put it to work on an incident triage workflow in Box. Jev doesn't generate text. You send it state plus typed questions and it returns booleans with probabilities, enums, and scores in a single pass, so there's no output to parse and…

Added to Jevfast

How it uses Jev

The workflow classifies a Box incident and sends low-confidence routing to human review.

App workflow

  1. Build state

    The demo pulls a Box incident report and builds the state used for triage.

  2. Ask triage questions

    Jev evaluates whether the incident is customer-facing and how severe it is.

  3. Check confidence

    The app compares the returned confidence estimate with its threshold.

  4. Route and record

    Confident results route to Escalate or Monitor and populate metadata; below-threshold results go to Review for a human.

Why it is interesting

The workflow uses confidence to decide when an incident should leave automatic routing and reach a human. That makes the review boundary part of the product flow, rather than hiding uncertainty in a folder choice.