How it works
Clarity Routing predicts the best queue/team/category for an incoming ticket/case by combining retrieval over your historical labeled items with a classification model.
Clarity is a routing decision engine, not a chatbot, and not a workflow automation suite.
Step 1: Seed with labeled history
You provide historical tickets/cases where the final outcome is known (e.g., final queue/team/category).
This creates a domain dataset with:
- Input text (subject/body/description)
- Final label (queue/team/category)
- Optional metadata (priority, channel, region, etc.)
Step 2: Retrieve similar examples
For each new intake item, Clarity:
- Uses the incoming text to find similar prior items
- Retrieves similar items from the same domain
These retrieved examples provide concrete, customer-specific context.
Step 3: Classify with example-grounded routing
Clarity uses the retrieved examples as structured context to predict:
- The best route (top-1)
- Ranked alternatives
- Confidence scores
If the primary provider is unavailable, Clarity can:
- Fail over to another provider, or
- Route requests to a private endpoint
Step 4: Operate safely (shadow mode first)
Clarity can run in shadow mode:
- Your current triage process remains the source of truth
- Clarity records recommendations and outcomes
- You measure improvements before enabling write-back
Outputs
Typical response fields:
- Predicted label (queue/team/category)
- Confidence score
- Ranked alternatives
- Explanation signals (e.g., top retrieved examples)
What Clarity is not
- Not a virtual agent
- Not a knowledge-base Q&A tool
- Not an automation platform