Revolutionising Sales with Intelligent Automation for the largest South African Short Term insurer
Insurance | Sales | Intelligent Automation
Case Study Details
We helped the largest short term insurer in South Africa shorten their non-life insurance quotation process from days to minutes by training a machine learning model to extract the relevant details required from competitors policy schedules.
Challenge
Our client wanted to reduce the time it took to provide its clients with short term insurance quotations and at the same time improve its operational efficiencies by automating much of the repetitive, error prone data capturing activities.
Solution
Lightblue trained a natural language understanding model and built the supporting pipeline to recognise and extract all the relevant data required from the majority of South African short term insurance provider’s policy schedules. This enabled out client to put a customer facing solution on their website to enable their customers to upload their existing competitor’s insurance schedules and get a firm quotation within minutes as opposed to days.
Approach
- Custom Ingestion Pipeline: Lightblue built a pipeline to ingest the policy schedules, identify them, split them into their relevant insurance line sections and pass this on to the trained model.
- Machine Learning Model: Lightblue trained a natural language understanding model to recognise and extract all the relevant entities and relationships for the majority of the customers competitor insurance policy
- Deployment: The final solution was deployed as an API to be consumed by our client’s customer facing application with a back end administration user interface to manage the ingestion pipeline.
Results
The implementation of our intelligent policy schedule ingestion solution yielded remarkable results:
- Operational Efficiencies: We were able to automate a task in minutes which would have taken one or more employees hours if not days to do.
- Quotation speed: We were able to shorten the time to quotation to minutes with significant increase in customer satisfaction.
Lessons Learnt
- Training Data: The importance of quality, representative training data is paramount to the performance of the model to be trained.
- Iterative Development: Adopting an iterative development approach, such as Agile, allows for flexibility and adaptability. Regular feedback cycles and incremental releases help address issues early and ensure the solution evolves in line with user needs.
Conclusion
This project underscores the transformative power of natural language understanding machine learning models in their ability to provide capabilities which were previously reserved for the human workforce.