by Daniel Lambert

Tackling artificial intelligence using architecture

Dec 19, 2018
Artificial IntelligenceFinancial Services IndustryTechnology Industry

CIOs know that they need to get involved with AI regarding their strategic initiatives, yet they are uncertain how to approach AI. An example from the financial services industry could show you how to develop, integrate and deploy AI in synchronization with the business strategies of your organization.

artificial intelligence brain machine learning digital transformation world networking
Credit: Getty Images

Artificial intelligence (‘AI’) is more and more sneaking up into our daily activities. Anyone using Google, Facebook or a Microsoft product knows this. It’s far from perfect, but it’s improving at a quick pace. Not every enterprise is using AI at the same pace. Has your organization started looking into using AI yet? Do you have any clue on how to tackle and implement AI in your organization? How should your enterprise and business architects examine AI? Where should they start? This article will try to answer these questions using a wealth management example.

What is artificial intelligence?

The first mention of artificial intelligence was about 60 years ago. AI has been defined in several ways. The10-minute video below, “What Is Artificial Intelligence Exactly?,” explains AI very well and elaborates on a few definitions:

I also find Wikipedia’s definition very appropriate:

Artificial intelligence (AI), sometimes called machine intelligence, is intelligence demonstrated by machines, in contrast to the natural intelligence displayed by humans and other animals. In computer science AI research is defined as the study of “intelligent agents”: any device that perceives its environment and takes actions that maximize its chance of successfully achieving its goals.”

Much of the recent enthusiasm about AI has been the consequence of developments in deep learning, which is based on learning data representations, called neural networks, as opposed to task-specific algorithms. Deep learning can be supervised, semi-supervised or unsupervised. Deep learning networks can now easily have over ten layers, with simulated neurons running into the millions, as mentioned in “The promise and challenge of the age of artificial intelligence.”

The deployment challenge

Not everyone has the deep pockets and the technical know-how of Google, Facebook or a Microsoft. Artificial Intelligence will most likely provide value, but its development, its implementation and its practical use is and will remain a real challenge for most enterprises, not to mention for most public organizations. Technical know-how and resources are scarce. Getting the right to, accessing and then analysing existing collected data will continue to be an issue in some circumstances. Finally, positive results from concrete artificial intelligence initiatives may prove longer to materialize then anticipated.

As mentioned by Andrew Ng, founder of Google Brain, in a recent article in Forbes:

“AI technology is exciting, but it is also immature. At the risk of sounding sacrilegious, AI technology in isolation is not useful. It needs a lot of customization to figure out exactly how it fits into your business concept. Doing that requires a broad understanding of your company and a reasonably deep understanding of AI. Exploiting the value of AI today requires a team that understands the business context and has cross-functional knowledge of things like how to fit AI into your hospital or how to use AI in your logistics network. Without cross-functional knowledge of how your business runs, it is difficult to customize AI appropriately to drive specific business results.”

Deploying artificial intelligence using architecture

As indicated by Raj Ramesh in this podcast about how business architecture can help leverage AI:

“Business architecture has a huge role to play in the future of organizations. There is no doubt that AI will be an integral part of the future business. Some of the key questions organizations ask related to the application of AI are things like “Where do we start?” “How do we mature the capabilities that will enhance our competitive advantage?” These are questions that business architects will help to answer when they map business strategy all the way to execution.

Enterprise and business architects are also becoming instrumental in designing future scenarios of these organizations using AI among others. Building and deploying AI applications cannot be executed with a chaotic approach. It is impossible to know where to start and make sense of AI without a rigorous business-oriented architecture beforehand. Business and enterprise architects must comprehend the appropriate information, value streams, capabilities, applications and processes that will be impacted by AI.

An example: artificial intelligence in wealth management

There are at least five ways AI is currently disrupting the financial services industry, as shown here:

  1. Investments through robot advisors will soon almost eliminate financial advisors;
  2. Chat bots backed by conversational AI abilities will soon enable customer engagement;
  3. Artificial intelligence has recently started to reduce false positives in fraud detection and risk management;
  4. AI will eventually be able to “learn”, remember, and comply with all applicable laws for regulatory compliance; and finally
  5. AI may soon be able to predict the price of stocks and market movements potentially turning upside down wealth management.

Let’s examine more closely AI and Wealth Management. If the circumstances are right, it may be appropriate to tackle an ambitious project and try to replace an investment manager of a non-performing fund with instead artificial intelligence.

lambert ai 1 Daniel Lambert

Managing a fund is essentially about selecting financial instruments to trade as shown in figure 1 above. It includes the following value stages: 1- examine financial instruments available for a selected category, 2- select the evaluation criteria of the financial instruments, 3- evaluate all available financial instruments, 4- select the Quantity and Price for each Financial Instrument of the fund and 5- place orders of the selected financial instruments. There are 16 capabilities in total enabling this value stream. 5 business capabilities enable the “Examine Financial Instruments Available for Selected Category” value stage. 5 capabilities enable the ‘Select Evaluation Criteria of Financial Instruments’ value stage. 3 capabilities enable the ‘Evaluate all Financial Instruments’ value stage. 2 capabilities enable the ‘Select Quantity and Price for Each Portfolio Financial Instrument’ value stage. Finally, 5 more capabilities enable the ‘Place Orders of Selected Financial Instruments’ value stage. Also note that some enabling capabilities enable more then 1 value stages.

lambert ai 2 Daniel Lambert

As shown at the bottom of figure 2 above, three critical enabling capabilities of the “Select Financial Instruments to Trade” value stream need to perform very well for AI to perform as well as the top quartile of the other competing funds in the selected category. The first problematic capability is ‘Financial Instrument Pattern Analysis’. The second one is the ‘Financial Instrument Valuation’ capability. Finally, the third problematic capability is the ‘Portfolio Allocation Determination’ capability.

Applications, processes and requirements related to these three capabilities will need to be examined in detail to complete the design of this wealth management artificial intelligence strategic initiative. Information concepts and databases related to the same three capabilities will also need to be studied.

lambert ai 3 Daniel Lambert

Enterprise and business architects can also elaborate and evaluate various scenarios to deliver portfolio management using artificial intelligence. Table 1 above describes briefly 4 possible scenarios. There could obviously be more. Financial analysis, impact analyses, and risk analysis will need to be completed for each scenario. Due to the high amount of capital involved and the uncertain timeframe of delivering a positive outcome, postponing the artificial intelligence initiative and stay ‘as is’ should not be excluded among the possible scenarios.

All artificial intelligence initiatives should always be examined in a similar systemic business and enterprise architecture approach, where the cross-functional knowledge of how your business runs is used to customize AI appropriately to drive specific business results. This will increase your odds of delivering a successful AI initiative in synchronization with the business strategies of your organization.