AI for labeling and routing
Compare AI that labels tickets, routes requests, and tags documents, from large language models to small, fast decision models.
Evaluation requested
No results yet.
Coverage
9 use cases
Methodology
Give every system the same items and the same list of categories, including none. Check answers against labels reviewed by people who do the work, and review every item where systems disagree. Report how often each system is right when it is confident, whether answers change when the list is reordered, and cost and response time per million items.
Categories
Label a ticket
Choose the one category that fits a ticket from a fixed list.
Route a request
Send each request to the team or queue that should handle it.
Say when nothing fits
Choose none when no category applies, instead of forcing a match.
Know when to ask a person
Be right when confident, and leave uncertain items for a person.
Give the same answer every time
Keep the answer when the categories are reordered or the input is reformatted.
Read long threads
Find the deciding detail in long tickets and comment histories.
Tag a document
Answer many yes-or-no tags about one document, such as a company filing.
Grade work on a scale
Score an answer against a rubric the way an expert reviewer would.
Check before acting
Decide whether a reply or tool call follows the rules before it goes out.