Part two of a talk by Tavio Roxo, CEO of OWLS™️ Software, at the 44th OESAI (Organisation of Eastern & Southern Africa Insurers) Conference in Zanzibar.
In part one, which was published in the September issue of COVER, I unpacked the challenges facing the insurance industry in our quest to improve the image of the industry while creating trust and confidence with our clients. The demise of Blackberry, Nokia, and Sony/Ericson, as described last month, sets the scene.
The architectural attributes of a tech stack for the future of insurance, now:
Firstly, any technology choice must be cloud-based. Sometimes referred to and native cloud, or full cloud. Unless the technology is full cloud, there will always be friction between where you want to go as a business and the structural limitation of your technology stack. Cloud-based technologies have many attributes that lend themselves better for use in a digital journey.
They provide full scalability, maximum availability and accessibility and are secure. You can spin up additional hardware resources easily with little to zero lead time. Due to the hardware array of a cloud setup, you can create full availability and accessibility. These environments are backed up across multiple pieces of hardware across multiple jurisdictions. This, by implication, can provide full back up If configured correctly, using the cloud allows for securing the environment through one access point for example. Also, with multiple firewalls, which are being updated hourly to mitigate against worldwide threats identified, you can have some comfort that threat levels have been mitigated. Gone are the days of doing this in-house with a server room in the basement… On premise install it was referred to, I think. That old school. That is Nokia stories…
To illustrate cloud-based technology VS older legacy technologies I will use an example.
From 2015 Mercedes S500 had the technology and did process sufficient data points to allow the car to keep the correct following distance from the car in front of it (adaptive cruise) whilst keeping in the lane (Disetronic +). This allowed for what is called “partial autonomous driving.”
Mercedes today however does not have FULL autonomous driving. Why? Mercedes was collecting those hundreds of thousands of data points every minute. The limitation, however, was that due to the legacy software stack they captured and stored those data points locally on the car… they were using an older tech stack (non-cloud) and although that was reliable it was limited in this fundamental way.
Mercedes had a balance sheet of 88 billion Euro in 2015. They could have pivoted. But did not. Tesla today however has full autonomous driving. You can get in, punch in your destination and the car will navigate you to your destination without you touching the accelerator or steering wheel. Tesla in 2015 had a balance sheet of 8 billion dollars (approximately one tenth that of Mercedes at the time). So how does Tesla get it right and Merc not?
There are various contributing factors to Tesla being able to beat Mercedes to autonomous driving, but primarily…Tesla is a completely different tech stack to that of Mercedes. A full cloud or native cloud tech stack. Which means it works differently. Autonomous vehicles generate as much as 40 terabytes of data an hour. This is from the cameras, radar, and other hundreds of sensors. In context… This is equivalent to 3000 years of iPhone use. Every hour. It’s mad. – There is a whole other discussion around data and storage… But they only gave me 30 minutes…
The significant difference of course being that where the Mercedes was collecting terabytes of data points, it was storing them locally (or in later years sent limited info). The Tesla, however, being a native cloud, was uploading this information from the get-go into a central computer that currently boasts a memory and processing bandwidth of two terabytes per second. An absolute beast. In fact, Tesla announced in August 2022 that it has the 7th largest Supercomputer in the world.
And it needs it… You see, all the Tesla cars are feeding all their data points to Tesla’s Mothership Supercomputer called Dojo and with these troves of data, Dojo is computing hundreds of billions of variations of different road traffic behaviours. Using Machine Learning and Neural Networks Tesla is forming a better picture every minute on how to ensure that your Tesla car drives safer. It will send all the updates to you at night while you sleep… It truly is a different world.
OWLS™ Insurance Software
Proud providers to insurance companies, UMA’s,
administrators, intermediaries and financial services companies.
The second learning, wherever possible, use a technology solution that allows you to perform all the functions end-to-end, for both life and non-life business, in the same place on the same system or tech stack. If not possible, then a recognition in your business that that is where you would want to go and limit the number of technologies.
As illustrated above, it is sub-optimal if you must step out of one technology into another to perform a business function. You see if you are performing all the functions in your business for both life and non-life on the same system you can record, store, use and compute all these data points. You are gathering information consistently. Like a Tesla. To a central place. If there are any pricing specialist in the room, they will be rubbing their hands.
We must perform all functions centrally: Quoting, Underwriting, Onboarding, Fulfilment, Servicing, Premium collection, and automated reconciliation, Claims management, Reporting, and Posting of financial transactions.
If you have this set up. Then you can immediately see this alignment. You can drive product rules immediately into this architecture. You can make changes on the fly. You can create products. You can distribute it. You can change it. You have the flexibility. Importantly. You can sell and distribute digitally. And you can administrate digitally.
But where the real magic happens… is because you have everything in one place. In a normalized data base structure… you can add automation. Automation in our insurance world is incredible. You gain orders of magnitude in efficiencies.
Then Thirdly, an appreciation that sophisticated IT skills are in shortage in our industry and on our continent. So, make it simple. Use a system which is, mostly, configuration vs development. When developing a piece of functionality for the first time a system developer can choose to hard code the functionality item. Or develop the item in such an agnostic way that it is configurable at a later stage by other business use cases.
Developing with later configurations permutations in mind is time consuming. In fact, much more time consuming. Depending on the requirement it could be 8X more challenging and time consuming. The advantage being that if done correctly at the outset, it significantly reduces the amount of time later to deploy the function. By the same factor 8X. By way of example imagine a large Distribution board. The electrical board that is found in most homes. The DB board is where all the circuits meet. In the DB board you can flick up a switch to switch on your stove. That is a configuration in an IT world.
Development on the other hand is the equivalent of putting that switch into the DB board. You must run a wire from your stove, through the kitchen wall, and chafe one wall. It comes with a lot of effort and with risk. You see whilst they are installing that stove switch someone touches another wire in the roof and this causes a short. So now we have tested the various lines to find the issue. So, favour a configuration toolkit. It will de-risk the project. It will also allow for the training of staff on an already existing piece of functionality. But the biggest attribute is that it works.
On our continent. We desperately need to train up, educate and build up significant internal skills within our organizations so we can create a sustainable insurance ecosystem for us all. I mean, if we have what we have today at a 2% insurance penetration rate on the continent in terms of skills and resources, and we are targeting the 14% penetration rate. That is a growth of 5X. We must rely on heavy lifting technology for some of it, but we also must rely on our people.
I am an African. Our business is African. If WE can build and deploy tech… against all odds…while everyone told us, we could not. Then WE all can.