Future of the Financial sector in the age of cloud — Google Next Panel Discussion
Today I attended a panel discussion about the future of finance in a world with cloud and data analytics. Some very interesting attention points sprung up, as well as some insights into how the world of finance will have to adapt the coming years. The panel consisted of speakers from ING, HSBC, HyperLedger and Blackrock and took place at Google NEXT 2018 in London.
On the topic of migrating legacy to the cloud
The point that was stressed the most, is that in order to be successful, it is important to introduce change gradually. Big bang changes may disrupt business, and although the impact on your bottom line may be dramatic, the worst change will be one that you can’t measure: you introduce aversion to change.
It is hard to convince the average Joe to change the way he has been working for years, it is even harder if he has had a bad experience in the past when this change was forced upon him.
When adopting the cloud, make sure to introduce change in a way that is constantly monitored, pay attention to the three key aspects: financial, human satisfaction and functionality.
When changing the technology, don’t try to replace existing processes exactly the way they are being done right now. Try to embrace the newest developments and focus on exploiting all the strong points. (e.g. when modelling on the cloud, don’t collaborate using Windows fileshares, use the accompanying toolboxes). This holds true for the typical Microsoft office workflow. Should you adopt GSuite, stop mailing attachments, but share documents on GSuite.
On the topic of adopting big data
HSBC has made one point very clear: on-prem big data doesn’t work. The upfront cost for setting up a cluster is high and technology changes so fast that nowadays you can’t expect clusters to stick around long enough to realize proper return-on-investment. On top of that, you need to train administrators to manage your stuff, you need to introduce role management, keep on track of ever changing frameworks and patch them, update them, maintain them …
It is a mess. The fallacy that cloud is unsafe is costing lots of businesses valuable time and those that embrace the cloud first have lower risk of missing important optimization opportunities.
Side note: Not in the panel discussion, but elsewhere on Google Next, I heard that a typical company that has embraced the cloud has shifted about 10% of their workload on the cloud. This is very important in setting up processes and developing the correct mindset, but it means that the competition still be competitive for now. But when those companies go to 30–40%, those who didn’t embrace cloud will be left out and have a very hard time catching up.
On the topic of IOT and Blockchain.
The consensus is that both Internet of Things and Blockchain have exited the stage of will they/won’t they and now the question remains when and how. Most agree that they will have a very noticeable effect on the way things are being done, but they also agree that it is impossible to predict how exactly. We can only hope that companies are sufficiently agile by employing cloud techniques to adapt to demanding and fast-paced innovations. IOT will greatly accelerate the inflow of data volume.
Both technologies will force companies to become more user centric, since the data will exist to make that happen. Those who don’t, will lose customers to those that are leveraging these technologies in the long run.
On the topic of Artificial Intelligence
All agree that Machine Learning is changing the way they do business, but they don’t want to use the word Artificial intelligence. There is no cognitive system, no interactive learning going on. ML is where it is at right now. We may also not forget that we are handling the money of the clients, mistakes can have very big consequences. Typically ML processes should be setup so human intervention still happens near the end, so that dangerous transactions or predictions that may impact the customer directly are spotted in advance.
Another approach is doing what NASA does, let multiple teams develop the same model, and only trust the prediction when all models more or less agree.
Note: This is not ensemble learning, this is letting different human teams work independently and then letting their models form a committee.
Very important keypoints were made, and it is clear that the world of finance will need to embrace these new technologies. The end user will benefit, the impact on the big companies remains unclear.