What it does
It reports per-ticket latency with p50 and p95, and accuracy overall against accuracy above a movable confidence threshold, so the abstain rate stays visible.
Playable demos & bots
Jev × BANKING77 — live classification demo
A single-page app that classifies BANKING77 support queries with Jev and streams the results one ticket at a time.
Bookmark: Jev × BANKING77 — live classification demo
Keep this in your collection.
Leave a noteWhat would you try with this?
: Jev × BANKING77 — live classification demo
Only you can see your notes.
Recording unavailable. Open the experiment ↗
How you can use it
To try this, gather sample customer questions and mark each one with its correct category. Your developer can set up and run this ready-made demo application for your team. They will just need an access key from TypeSafe to connect the app to the Jev model.
Your developer can swap in your own list of questions and categories. You can then use the on-screen slider to see how well the system sorts messages. Moving the slider shows which clear answers pass through and which uncertain ones wait for a person.