Platform

We have committed all our experience in and knowledge of insurances to building the best platform we can imagine you configure, value and administrate all your insurance products in the best way possible.

If you’d like to learn more about how we did it, and what the resulting platform looks like, you can find out about it below!

Target Architecture

TrustFlow will form the back-bone of all of your IT infrastructure. Below, we give an infographic showing an indication of what the architecture could look like, what modules are offered by TrustFlow and how they integrate with the rest of your software.

Insert architecture overview here

Back-end

The back-end constitutes the core of our solution, and this is where the clever solutions are employed that ensures with TrustFlow, you’ll be able to configure, store and value all your insurance products accurately and efficiently.

  • In Trustflow, data is stored and processed per cover. Data structure depends on the type of cover and is customisable based on your requirements. This data is stored in relational databases, supporting all ETL functionalities.

  • Depending on your preferences, assumptions can be stored in either a relational database or using a file system in which assumptions are stored in files in a specific format (e.g. Excel/csv files with a fixed structure). New assumptions can be easily added, which we always validate. We store old assumptions in reference to calculations that have been done with them.

  • The solution allows for the valuation of the options and guarantees using Replicating Portfolios or Monte Carlo Simulations. Both methodologies can be used to calculate the intrinsic and time value.

    The risk margin is calculated per policy (applying the different shocks) and then aggregated to HRG and portfolio level to include diversification effects between policies and products. The risk margin for a policy can be calculated using risk drivers to project the capital requirements for the different risks into the future or using a full calculation to exactly calculate the required capital at each point in time (which is more computationally heavy). Full calculation of the required capital at each point in time is also incorporated.

    When setting up a run the user can specify whether calculations are performed per cover or on a more aggregated level where covers with similar risks (e.g. age, gender, duration, etc.) are combined into a single record to speed up the calculation. Furthermore, cashflows can be calculated on a monthly/annual basis. Furthermore, the user can specifiy on which level the data should be stored (e.g. per policy, per HRG, etc.)

  • We provide benchmarks which reproduce the results of the system exactly. So in order to trace back a certain result the user can automatically import all data in the benchmark from the actual calculation and then take a look at the results and see all the intermediate results/formulas for the different parts of the calculation such as:

    - The number of policies per Markov State

    - Generation of the cash flows for each of the SCR scenarios

    - Calculation of the reserves

    Furthermore, when a calculation is performed all the intermediate results as stated in the list above can also written to the output.

    The user cannot directly see/interact with the software that is being run, our experience is that it is easiest to provide transparancy for the formulas using our excel benchmark technique rather than looking at source code.

  • A variety of analysis are available, and the types of analyses can be expanded upon request or by yourself. The following are included:

    - Cash flow, amount of policies per markov state, Best Estimate, SCR and RM calculations

    - IFRS17 calculations

    - Calculation of the Value New Business

    - Calculation of the cash flows and best estimate values for an individual policy (this is used for debugging purposes)

    - Projection of the best estimate

    For the above the effects of reinsurance can also be included in the analysis if required.

  • Assets which are currently supported in the application are:

    - Equity (type1, type 2, strategic)

    - Government bonds

    - Corporate bonds

    - Deposits

    - Interest rate swaps

    - Interest rate swaptions

    - Real estate

  • Products which are supported out of the box are:

    - Term Life

    - Funeral

    - Annuity

    - Unit Linked

    - Pension DB and DC

    - Individual Disability

    Other products can easily added by us upon request, or by yourself.

    For valuation runs, we allow flexibility in that manual corrections can be added before starting a run. It is possible to add a manual reserve or an additional cash flow projection to make manual corrections.

    Furthermore, aside from analyses supported out of the box, the flexibility of our solution allows you to calculate any desired custom metrics or perform any desired analyses.

  • The following reinsurance contracts are already supported in the solution: Quota Share, Excess of Loss.

Front-End

This is the part of our solution your employees and customers will ultimately interact with, so we are committed to making it as easy-to-use as possible. Below, we highlight how to each role can interact with our system, and what functionalities they can enjoy.

  • We offer our solution as a mobile- and desktop compatible browser application.

  • Standardly, we offer the following functionalities for customers:

    • Fully digital application process.

    • Digital portal in which personal information an d policies can be managed.

    • An as digitized as possible claims process.

    • Overview of full mutation history.

  • For insurance employees (depending on permissions, .i.e. Role Based Access Control (RBAC)), we offer the following functionalities standard:

    • Case management.

    • Policy and customer management.

    • Digitized claims processing.

    • Automated collection and disbursement.

    • Digital product management.

    • Full audit trail of all mutations.

    • Complete and real-time overview of all financial metrics through a dashboard.

Workflow and Runmanager

We understand efficiently and accurately being able to perform calculation runs is of the utmost importance to any insurer. That is why we have ensured our workflows and runs are easy to set up and intuitive to use and interpret.

  • It is possible to schedule runs to run periodically (e.g. each month) or at a specific moment (e.g. overnight runs when infrastructure costs are lower).

    The system can run multiple different calculations in parallel. For example, an AoC can be calculated automatically.

  • The metadata includes, the start/end time of the calculation, the version of the software used, a reference to the input that was used for the calculation (this input can never be changed), the person who started the calculation, the server that was used to perform the calculation and also a log of the entire calculation flow.

  • See ‘Back-End - Results Analysis’.

API’s and other couplings

Any modern software system should easily integrate with other systems for automation and efficiency. Below, we highlight how we ensure and facilitate that at TrustFlow.

  • We make use of GraphQL in order for the front-end services to collect and mutate information. Furthermore, the Batch API allows mutations through files. If desired and possible, we facilitate the integration of any API-enabled applications you want integrated.

  • We aim to integrate TrustFlow as well as possible into your business logic as desired. For example, all data is written to a database which can be linked to PowerBI. We would love to discuss any other ways to best integrate TrustFlow into your business logic!

Efficiency and scalability

Actuarial computations can be burdensome and take a long time. Below, we highlight how we ensure efficient computing through our systems, and give a glimpse of the expected run times. Also, we'd like to explain how our system easily scales with your organization.

  • A couple of solutions are used to achieve high performance:

    • The solution runs all the cash flow calculations in parallel

    • Makes heavy use of caching for calculations that are done multiple times when projecting the cash flows (such as tariff calculations for lapse, premium free etc.)

    • We make use of performance profilers to ensure that the model runs efficiently

    • The solution runs on the Azure Batch service which automatically scales depending on the number of calculations being performed. When no calculations are performed no costs are incurred for the infrastructure

    • For generating the cash flows under various economic scenarios we make use of the replicating portfolios. We pioneered this technique and were the first to use it in a commercial setting. In order to construct the replicating portfolio we run the model once using 10,000 economic test scenarios and then fit the resulting cash flow results to a portfolio of assets which generate the same cash flows for all these scenarios. 5000 of these scenarios were used for fitting the cash flows to a portfolio and the remaining 5000 were used to validate the fit. This technique allows for generating the cash flows in real-time, solvency levels, providing an overview of the risk indicators and providing dashboarding with a traffic light system to warn the insurer if the solvency is in danger. The entire process of generating this dashboard was done daily and completely automated (together with a real time connection with bloomberg and the asset manager).

  • A complete SII run in HiThere incorporates both the valuation of the assets and the liabilities including calculation of the SCR and RM. The run time depends on a number of factors such as: the number and type of covers (lifelong or temporary covers), the granularity of the calculation (monthly/yearly cashflows), the selected infrastructure (number of parallel cores).

    As an indication, for 500,000 policies a complete SII calculation on a medium infrastructure with monthly cashflows had a run time of around 3 hours. Given our cloud infrastructure, the user can perform multiple of these runs simultaneously or schedule them to run for example overnight.

  • We have good experience hosting the solution on the Azure platform (although it can run on any platform), more specifically the Azure Batch service is used for performing the calculations. Here, you can run as many calculations in parallel as you wish and can scale endlessly in this regard. In addition, if no calculations are running then no costs are incurred for the calculation server.

Hosting

We facilitate both on-premise hosting as well as a completely cloud-based systems. Below, we explain how we can ensure our system is deployed according to your needs and wishes!

  • The system can be deployed in on-premise environments.

  • In the cloud, TrustFlow can be configured in any environment as desired, such as Azure or AWS. We have particularly good experience with Azure, in which case it is hosted in thee West Europe Region of Azure located in the Netherlands.

Security and Compliance

We know you take the greatest care to ensure the security of data of your customers, and IT infrastructure is one of the most important places to do so. Therefore, we employ all the necessary measures to guarantee your data is securely stored and handled in our system.

  • All software is written following strict code security, quality and testing guidelines. Furthermore, we employ the following security practices:

    • Encryption at rest and in-transit

    • Vulnerability scanning

    • Automated unit testing for security

    • All logic is back-end contained, such that security cannot be circumvented through front-end.

    • Integration with existing identity service provider

    • OTAP

    • Minimum access policy on infrastructure level.

  • A permission layer governs access control for employees, ensuring data can be retrieved by intended users only.

  • We minimize the downtime by creating redudancies in our infrastructure, so there can be multiple environments hosted in different availability zones in the infrastructure. Finally, as a last backup we have made some agreements with our customer that we perform the calculations on our own on-premise infrastructure and share the results as a worst case backup, this is something that we have never had to use but can be arranged.

Employed technologies

To build our platform, we have made use of various state-of-the-art technologies. Below, you can find out which ones.

  • Programming languages: Javascript, HTML, CSS.

    Framework: React.js

  • Programming language: Kotlin

    API technology: GraphQL