Mine tickets for training data
Connect
SPIRITT
Wit.aiWit.ai
Language models for your product
SPIRITT builds Wit.ai language understanding the way an NLU engineer would. It designs intents, entities and traits, writes and uploads labeled training utterances, tests parsing on real messages, detects languages, and exports backups before every change.
What you can do with Wit.ai.
Models designed from your data
Apps, intents, entities with roles and keywords, synonyms and traits, created from your real requests.
Training that keeps improving
Labeled utterances uploaded in batches, reviewed and pruned as the model learns.
Parsing in production
Messages parsed into intent, entities and traits, language detection, and versioned backups.
Works with your other apps.
Get started in three steps.
01
Open your workspace
Start a SPIRITT workspace for your language understanding work and describe what users say and what it should mean.
02
Connect Wit.ai
Connect Wit.ai, plus the sources of real user messages such as your help desk or chat logs.
03
Hand over the mission
Hand over the model. SPIRITT designs, trains, tests and maintains it.
Hand over a mission.
Choose a mission. Make it yours.
Put Wit.ai to work.
Language models for your product
Questions
Wit.ai, with SPIRITT.
01Can SPIRITT build a Wit.ai app?+
Yes. It creates apps and defines intents, entities, roles, keywords, synonyms and traits.
02Can it train the model?+
It uploads labeled training utterances, lists them for review and deletes bad samples.
03Can it use the model on real messages?+
Yes. It parses text into intent, entities and traits and can detect the language.
04Can it back up a model?+
It exports the app as a backup ZIP and lists versioned tags.
05Can it remove parts of a model?+
Entities, intents, keywords, synonyms, roles and whole apps can be deleted.
06How do I connect Wit.ai?+
Open a workspace, connect Wit.ai and your message sources, then describe what the model should understand.