$5.2B annual revenue (2020)
Global corporation with operations in 140 countries
1,500 offices and laboratories
78,000+ employees
Widely considered a world leader in TIC services
Verisk is a pioneer in data analytics, delivering innovative risk and investment analysis for the past 50 years. The company used Super.Classify to process large volumes of unstructured sales interaction data, illustrating how artificial intelligence (AI) can improve information accessibility—leading to more satisfied customers and revenue.
The company operates globally with activities in 34 countries and is publicly traded on the NASDAQ. Keen to put data to work to improve their buyer and customer experiences, Verisk sought automation expertise to scale data processing while adhering to stringent privacy requirements, in particular for recorded customer calls as well as text and documents. Struggling to process unstructured audio data, as well as maintain data privacy throughout the analytics process, Verisk turned to an AI-powered solution.
By building an intelligent chatbot that automatically responds to customer sales queries with the most relevant information, the company increased customer satisfaction and retention while freeing up human resources, accelerating customer service response times, and adhering to strict data privacy requirements.
Verisk selected Super.Classify, super.AI’s no-code solution for data classification, due to its end-to-end capabilities, from data processing to model deployment. Super.AI stood out in the search for a data processing partner for offering multi-modal data processing and active learning capabilities, so that the customer can assist with ML models, in this case to analyze their own sales calls.
The customer data processing team started by employing super.AI audio transcription data program to transcribe all sales call audio files into text. The second step was to use super.AI query intent matching data program, inputting search queries and possible intents for each query, producing an outcome file with score matching intent for each query, on a scale of 0-1 (no connection to 1 (perfect match).
In parallel, Verisk’s processed data was fed into an active learning AI algorithm automatically. Simultaneous data processing and training the active learning algorithm enabled Verisk to significantly reduce time to deployment of the ML model, while ensuring a high degree of quality within their cost parameters.
The solution delivered a significant increase in relevance of recommended responses to customer questions during sales calls, contributing to increased customer satisfaction, retention, and revenue.
Deployed project in # weeks from kickoff to production-ready application
Enabled Verisk internal analyst team to train AI model to maintain strict data privacy
Integrated via API to upload data programmatically and fetch results automatically
Transcribed more than ##### recorded audio sales calls
Processed over ##### files in just ######
Generated a library of over #### customer query response recommendations
The super.AI product suite offers no-code solutions for processing video, image, audio, text, and other unstructured data automatically using AI. Build artificial intelligence applications #X faster than alternatives thanks to pre-trained models. Guarantee 99%+ data accuracy using a combination of AI and human workers. Some real-world applications of AI in TIC services include:
Nameplate Extraction: Automatically extract information such as equipment serial numbers, model numbers, product expiration dates, and more from inspection images.
Automate Vehicle Damage: Automate vehicle damage detection and achieve a new standard of accuracy and speed at a lower cost.
General Damage Detection: Automate the detection of equipment or building damage such as surface cracks and water intrusion using inspection images.
Corrosion Monitoring and Detection: Automatically identify corrosion in boilers, pipes, storage tanks, vessels, and equipment using images and videos.
Product Defect Detection: Automate the detection of defects in products during manufacturing, such as cosmetic issues, poor welds, and assembly flaws.
Seismic Risk Assessments: Automate the analysis of seismic hazards and assess structural vulnerabilities.
Fuel System Inspections: Automate the processing of fuel system inspections for wear, damage, or leaks.
Equipment Inventory: Automate asset tagging and management by quickly transcribing the tags into your format of choice.