Actuarial Processes Revolutionized - Introduction to ActuAI

Machine Learning in Capital Modelling

March 3, 2020

ActuAI as Model Serving Infrastructure

July 27, 2020

The insurance industry is experiencing changes due to rapid advances in artificial intelligence (AI), data science and analytics. According to International Data Corporation, the expenditure on AI in the insurance industry will rise from 24 bln USD in 2019 to predicted 77.6 bln USD in 2022. The correct implementation of AI techniques will become the main competitive differentiator in the insurance industry in the next five years.

The investment in AI have been concentrating on chatbots, wearable devices, claim settlement tools and sales recommendation systems, however the insurance companies are starting to merge data science, actuarial and IT teams to bring value and automation to actuarial processes such as pricing, claim & loss prediction or capital modelling.

Nevertheless, the current pain is that it requires millions of $ to build, deploy and maintain AI solutions. Here is the short summary of the current biggest challenges to implement AI in actuarial departments:

HIRING DATA SCIENCE TEAM

Current high demand for data science talents and shortage on the job market.

DEPLOYMENT AND MAINTAINCE

Deployment of proof-of-concept to production can take months and engage numerous IT people.

COMPLEX AND REPETITIVE TASKS

Actuaries must rebuild models, test them and deploy in short frame of time due to regulatory and stakeholders' requirements.

EXPLAINABLITY

Insurance regulators and other stakeholders require explainability and do not accept black-box models.

Particularly in actuarial processes such as pricing, claim prediction or risk management there is a lack of skillful workforce, time resources and/or external software to utilize benefits of data science in insurance. That is why we invented ActuAI Platform – an enterprise software solution tailored to insurance data science applications.

What is ActuAI?

Quantee ActuAI Platform is an easy-to-use, secure and auditable cloud-based solution for actuarial departments in the insurance and reinsurance sectors. It is software designed by actuaries for actuaries to automate, enhance and accelerate traditional actuarial processes with automated machine learning algorithms, explainable AI and understandable visualizations.

How it differs from other external software and internal data science teams?

ActuAI is the first platform for multiple actuarial processes using true artificial intelligence to boost actuarial modelling.
ActuAI offers great flexibility that actuaries expect.
ActuAI allows you to utilize the benefits of automated machine learning effortlessly, saving much time on PoC and building models in production.

ActuAI is considered as end-to-end actuarial data science platform containing modules not only for non-life pricing but also for life/non-life lapse analysis, claim prediction and capital modelling. It follows the actuarial modelling pipeline involving raw dataset upload, a robust selection of data transformations and both benchmarking actuarial models as well as interpretable machine learning algorithms.

These algorithms are tailored to specific meanders of actuarial modelling, hardly available in open-source software such as exotic loss functions, exposure or offset variables and visualizations just for actuarial reporting purposes.

Our team (mostly consisting of qualified actuaries) works with an aim to build software allowing for great flexibility. Therefore, the modelling can be:

Performed in Auto Mode, using our Bayesian optimization technique for selecting proper data transformation and models in a completely automated way.

Performed in Manual Mode so that actuaries have full control of the process and get insights from each step.

Automated thanks to a vast selection of built-in data transformation and estimators dedicated to actuarial processes in both regression and classification tasks including frequency, severity and survival modelling.

Experienced actuaries can write their data transformation or estimator in R/Python, upload it to our platform and still benefit from other steps, visualizations as well as the whole model serving infrastructure.

Done in an intuitive front-end with perfect UI/UX design.

Programmatic access to each step is available through REST API so that the ActuAI can be integrated with other products, processes or even departments!