Data Intelligence product
Turn text into an answer you define
No training data
Resolve starts with the question and the answers, not a labeled dataset. It does not use Lake gold. Send a piece of text and get a choice or a score back from an API.
Write the answers. No labeled dataset and no data science team required.
Two kinds of answer
Pick one, or place it on a scale
Route a ticket
A customer writes “I was charged twice.” You defined Billing as “a charge, a refund, or an invoice” and Support as “something is broken.” Resolve returns Billing, with the probability on Support still visible, so a queue can send the message to the right team before a person opens it.
Score how urgent it is
The same inbox can ask a second question: how fast does someone need to respond? Levels run from “can wait” to “needs someone now.” The score can fall between two levels. Your app compares confidence to an optional cutoff and leaves the unsure ones for a person.
What you set up
01
Write the question
Name the Resolve and describe what you want decided. One question per Resolve. The type stays fixed after you create it.
02
Define the answers
Choice: 2 to 255 named answers, each with a short note on when to pick it. Score: 2 to 10 levels, written from low to high.
03
Send text
Issue a key on the API tab and POST the message, email, or note. You get the answer, a confidence, and the probability on every option.
The cutoff is optional. Leave it blank and every call returns an answer. Set 0.8 and your application can treat anything below that as “ask a person.” The API still returns the full choice or score and the probabilities either way.
Answer text without a training set
Write the question and the answers, deploy an API, and send the next ticket, email, or note whenever you need a choice or a score.