call_center_ai · finance · workflow
Verisk scales customer sales call processing with super.AI NLP and active learning
Verisk needed a scalable, privacy-compliant solution to process large volumes of customer sales call recordings and feed the results into an active learning algorithm, with the goal of improving recommendation relevance and increasing customer satisfaction and retention.
How it works
Common implementation structure
How this type of workflow is generally built, generalized across documented cases — not tied to any one vendor's stack. Click any stage to read what happens there. Specific products that implement these stages appear in “Tools commonly seen” below.
Stage 1 · Audio transcription of sales calls
Sales call audio files are transcribed into text using an audio transcription data program.
Tools used
super.AI NLP solutionaudio transcription data programquery intent matching data programactive learning algorithm
Outcome
By running data processing and active learning training in parallel, Verisk significantly reduced ML model deployment timelines while maintaining quality within cost parameters, and significantly increased the relevance of recommendations for customer queries during sales calls, driving higher customer satisfaction, retention, and revenue.
Results
Time savedsignificantly reduce timelines for the deployment of the ML model
Cost replacedincreasing customer satisfaction, retention and revenue
Grounding & classification
Source type: vendor customer story
25 fields verified against source quotes.
data extractionpredictive analyticsspeech to textcall recordinghuman review describednamed customerproduction runtime claimedsource backedtools describedworkflow describedfinancial servicesinsuranceaccuracy improvementcustomer satisfactioncycle time reductionvendor customer storycall center aisales opsextract classify route