Evolutionary Path – Six Pillars of the Evolutionary Approach

The evolutionary approach for Artificial Intelligence focuses on evolving your business by making existing products, service and business processes better. In the evolutionary approach we often observe that AI can carry out tasks cheaper than traditional means. To discover the right opportunity, it is best to start with your existing business practice and products. In our discussions with companies, we have worked on six fields to analyze the status-quo and uncover potential areas of AI improvements:

Fig. 10.2 Six pillars of evolutionary approach discovery

Let us provide some insight on tackling each area and devise the right application for your organization.

Your Company Data Streams

Big Data has shaken up a lot of organizations and created an awareness for the value of data. Although most companies we have worked with have engaged in Big Data projects, we are surprised to find that most organization are not aware of the plethora of data they operate on, where the data is stored and what costs are involved in creating, storing, and managing the data. The quality of the data is often in ‘suboptimal shape’, to put it nicely. When engaging with companies we often start by listing the data sources inside the company and the storage. To name a few: E-Mails, customer data, financial data, suppliers data, interaction data, transaction data, sales data, product lifecycle data. Thereafter we examine the data being used or generated in your products and service. Often all this data is stored in databases, data warehouses or proprietary software systems. Lastly we think of new data sets that can be created by adding additional data sources or engaging sensors.

In correspondence with this practice, we then carry out two exercises for each data set, with the goal of understanding the impact. We carry this out with a diverse group of experts from the company. If possible ?—? we recommend including a data scientist. Furthermore, do not forget to consider outside data sources that can be used to augment and enhance your company data. Here are the two exercises that you should conduct:

  • What decisions can be derived from each data set – this focuses on the primary usage of the data, but we also brainstorm how else this data could be used.
  • What efforts are needed to create, store and process each data set ?—? here we look at the total cost on how this data is being managed and used. Companies are often not aware how much effort is behind creating their own data flows.

Now let’s apply this using a tool that all organizations have at their disposal: your customers data. Think of your own organization. Are you happy with your customer data? Is it good quality and are you utilizing it effectively? Our findings conclude that for most organizations this is a huge potential to be exploited. Here are some examples of questions you might ask:

  • Who are your top customers and what are their characteristics?
  • Which customers are on their way to becoming top customers and what are the patterns that define this?
  • What is the next predicted purchase for this customer and what are the signs indicating this?
  • What customers are in the process of leaving your company for your competitor and how do you detect this?
  • What products are your customers purchasing and what products are reaching the end of their lifecycle soon?
  • Based on the communication and behavior, how happy are customers with your products and services, how loyal are they to your brand?

By crunching the numbers, understanding the patterns and predicting the trends your team of experts can figure out the answer to these questions. The data scientist can then utilize these insights and develop complex algorithms for your programming team to input into algorithms. It is even beneficial to carry this out manually as it can help you predict behavior and allows you to make better decision on product management, marketing, customer support, and many areas that consume most of your company’s budget. Most notably, this is where artificial intelligence can really make a difference. Neural networks are good at identifying complex patterns and making good predictions. AI is a great tool for helping your company understand your data streams, make sense of them, and allow you to make better decisions based on this data. We recommend running experiments on some of the data with the most leverage. By running those experiments you will learn how to engage with artificial intelligence as well as the pitfalls in quality of your data, confidence level, training the neural networks and you build up a new competence in an area, which will be a critical competitive factor in your company’s future.

Your role in all of this will be to identify the right areas where it makes sense to engage and develop your AI project. Then to find the right AI tools to help you recognize and understand the data, and finally, perhaps most importantly, your job is to find the right business model to act upon this, as we have learned in the first section of the second part of this book.

Your Software

Wherever you find data, software is not far. It is used to manage the data flows, to store the data, and to act upon it. Therefore another approach to discovering your AI potential is to review your software. You can do so by making a list of the software that is deployed in your organization. Often your software acts upon the same data, so it makes sense to also review what software is working together on data sets. Again ask yourself some question on each of the software you are using:

  • What data is your software creating or acting upon and what decisions do you take on that data?
  • What additional decisions would you like to take based on the data?
  • What efforts are needed to derive those decision and how can they be taken more intelligently?

Then rank the software and corresponding data by most impact both in terms of cost savings and also in terms of new insights and better decision to be taken.

Let’s take your HR software as an example. It lists your employees, their performance and feedback, their time spent working, vacationing, just to name a few. So what are the questions should you ask yourself?

  • Am I hiring the right people?
  • Which people are going to leave soon?
  • How is the motivation in my organization?
  • How are the managers performing?
  • What teams are creating most value and why?

Again, these are just a few examples. Your role in this will be to identify the right areas for your organization to engage, generate the data sources, enable AI to understand the patterns and find ways to engage that provide the most value to your organization. A number of software systems are being enhanced by AI from its creators or new AI based tools are available. It makes sense to do some research before you engage.

Your Processes

Most of your organization is likely to work on business processes on a daily, weekly, monthly, quarterly, or yearly basis. Make a list of your business processes, both primary and secondary. By now patterns should emerge as you dive deeper through each section.

Ask yourselves the following questions for each process:

  • What decisions are you taking based on this process?
  • How much effort is it to conduct the process, what are the total costs?
  • Could this process be automated? Could parts of the process be automated?
  • What insights could be taken on the data generated in the process?

Now rank your findings in terms of business impact. What are the biggest problems you would like to solve? Where do you expect to earn the greatest gain? Where can you significantly save costs? What is your ability to execute on these areas? There is no standard answer to these questions, we recognize that this is quite individual to each organization and this is your value in an artificial intelligence world ?— ?to translate the new capabilities into your business.

Another area is to look for modern software suites that are business process focused. Companies such as Atlassian, Salesforce, SAP, IBM, Adobe are only a few well-known players in the market. A whole new ecosystem of enterprise software has AI-capabilities built in right from the start.

Fig. 10.3 TopBots Enterprise AI landscape 2017

Your Interactions

AI is showing promising results in enhancing human and company relations, such as translating languages in real time from human to machine (bot). As well as customer service automation and bot-to-bot interactions (for business-to-business commerce). This is a prosperous area for your business to benefit from as it is expanding rapidly. List all your company’s interactions and ask yourself the following questions:

  • What decisions are being taken according to the interaction?
  • How much effort is it to engage in these interactions and what are the total costs?
  • Could part of the interaction be automated?

Again rank your findings in order of importance to your organization and identify the areas of biggest value. Note that interactions based on text, speech and image have been among the early areas of progress for AI. So, there are a lot of services and software solutions available to plug-into your enterprise to assist in these processes.

One example are chatbots or voice assistants like Amazon Echo, Apple Siri or Google Home. These AI-based conversational services allow you to automate parts of your customer interactions and thereby provide a personalized interface with a human-like interaction. Voice assistant services can be deployed through the available devices and chatbots can be integrated directly into your digital offerings or integrated into popular messenger platforms like Facebook Messenger, or WhatsApp.

The problem with this is that we often find that their experience is not perfect. Thus it often makes sense to deploy a combination of machine and human intelligence in the solution. If you take chatbots for example, the solutions can be a combination of machine and human where the machine introduces the chat and asks and responds to questions where it has a high confidence level. Questions more complex in nature are handed over to humans to carry on the interaction. This unburdening of customer support people can result in significant cost savings.

Fig. 10.4 Gradual takeover by AI from humans as the confidence level increases

Above is a model of joint deployment of human intelligence with artificial intelligence, where AI takes on more workload, as the confidence level increases.

Your Products & Services

The last area we will look at are your products and services. You can build AI into your products and service, because AI can get the job done cheaper meaning cost-savings for you, or because it can deliver new capabilities to your customers and ultimately increase the value of your products and services.

How AI can support your products and services can be based on simply adding mature AI areas, such as speech recognition, text to speech, conversation, image recognition. Look to the current set of capabilities of major AI platforms by IBM, Google, Amazon, Microsoft or look at the AI software landscape for capabilities to add to your products. You can achieve some significant results in making interfaces nice, more intelligent, adding better support capabilities and faster results for your users. Because of the proven areas, there is a lower development and integration time and it is easy to prototype and test the benefits before deploying the solution. Be aware though, that your competitors are also working on such solutions and because you are using standard AI services, it might not result in a long-term differentiation from your competitors.

As we covered in the first part of this book, an example of this might be the Netflix video service that created an Amazon Echo Skill that allows people to control their Netflix experience by simply using their voice. “Alexa, I want to watch my favorite Netflix series.” This can result in a complex multi-step service, which makes the experience so much more compelling. Amazon Echo turns the TV on, loads Netflix, tunes into the right person based on the voice profile, selects the favorite series, and launches the next episode. If you have deployed some smart-home solutions, it can also automatically lower the window blinds and turn down the lights without additional commands.

A more beneficial approach is to marry the USPs of your product and brand with AI capabilities. An excellent example of this is Amazon optimizing product suggestions based on AI or Google Translate improving their translation services based on Machine Learning or BMW offering smoother braking algorithms. Take your existing products and services and make them even better in areas that differentiate. To achieve this, you make lists in 3 areas:

  1. The use cases and values delivered
  2. The product / service features
  3. Your unique area of expertise and differentiation from the competition

For each item, list the data flows and think of how they can be made more intelligent and cost-effective. From this, you should also brainstorm what are new areas where you can deliver valuable use cases and offer new features. We recommend creating a systems map, listing on the left the stakeholders involved and on the right the end result. In order to visualize the process and data flows you draw boxes around them. Mark the areas where AI can help optimize the process. This will reveal what data is needed and how to get it?—? for example by deploying new sensors. This is not the last step though, you can deploy AI libraries in the right places to offer AI services and self-learning algorithms. You have to train these AI systems so think of training scenarios and getting training data. Then you go on extensive training phase to develop and train the artificial intelligence before deploying to your products and services.

Fig. 10.5 Example of a systems map from Google Venture SPRINT book

Reflecting again on the Salesforce CRM software. The company’s strength of offering software-as-a-service and combining many customer facing tasks in a unified platform offering was the basis for adding new value through their Artificial Intelligence offering’ Einstein.’ Traditional Salesforce CRM was good at ranking top customers based on revenue data for example, but Einstein allows you to understand why the customers are top. What are the characteristics of top customers, what other customers are on their path to become top customers? When will these customers buy again? Here Salesforce played on their differentiating strengths and they have made CRM much more intelligent and beneficial to their users. The result of this is it allows them to charge a premium. It is still early days, but we can foresee a time where companies will not buy a CRM system that is not based on artificial intelligence. It will be too costly for the companies, because business operations and decisions will not be as optimized as those where the companies deploy intelligent CRM systems.

This kind of Artificial Intelligence requires a deeper engagement by employees working on the core products and services. You also need to delve into unproven ways, which are likely to be more costly and risky. But the reward can be a higher differentiation of your products and services. Often this requires senior management commitment and a strong vision, if you wish to benefit from these capabilities early. If you are not an early mover, don’t worry?—? your competition will drive you to this space very soon, as we are entering the era of artificial intelligence.

Going through this process helps you identify areas where you can gain most benefits from AI. They are easy to identify and engage with and will kick off your AI journey. However, they are not sufficient to exploit the maximum benefits for your organization. Now we have explored this approach let’s explore an entirely different approach to identifying your AI potential.

Leave a Reply

Your email address will not be published. Required fields are marked *