In the previous section we saw that within operational readiness Data management is needed to optimize the use of data  and maximize the benefit to the organization. Instead of trying to optimize the multitude of data sources on their own we discussed that it would be better to optimize the business process management which creates the data we want to optimize. The process of optimizing and holistically combining business process management is being described as enterprise resource planning. That means that by introducing ERP software into our business, we automatically come one step closer to operational readiness.

Bigger companies have invested into ERP systems since the early 90ies and many of you have heard of companies like SAP.

However smaller companies, governmental institutions or hospitals still are not using ERPs at all.

A.I. readiness means digitalization and forcing yourself to have a holistic enterprise resource planning strategy, will automatically lead to dissolve data silos and data holes. These data holes are created when certain business units are not storing data or storing them in formats other business units cannot access (paper, fax, CD). If your initial idea is that any data created should be able to be accessible theoretically by anyone at anytime, then you will be able to create a solid ERP system.

For many companies, this does not start with IT at all. The first step of creating a good ERP system is to actually write down your existing business processes and standardize them. The first step is to understand your business workflows on paper. This understanding has to be shared by all your employees. Thus in order to start building a good ERP system, ERP consultants will come into your company and sit with your employees, in order to determine how work is done in your company and how data and information is flowing through your company. Corona has accelerated this process for many companies as many people went into home office and implicit business activities and implicit agreements had to be made explicit in order to lead remote teams. Many people realized how many implicit processes the company had, until they were home alone.

If we will look back in a few years to the time of Corona, we will see that the need for business process optimization and thus the common surge for ERP systems will have boosted many companies into operational A.I. readiness.

Over this week we followed Intel and dissected A.I. readiness into three dimensions: Transformational, Operational and Foundational readiness. Then we saw how Prof. Malone from MIT divided A.I. systems into four categories: tools, assistants, peers and managers. Then we started to look closer into operational readiness and especially data management and saw today that by introducing ERP systems in your company, you automatically reach a high level of operational A.I. readiness.

I hope that this chapter gave you a good overview into what A.I. readiness is and how you can achieve it.

As we saw in the previous section, operational readiness includes many areas of your business which are not related to IT or data at all. But for today we want to focus on data management. According to oracle,

Data management is the practice of collecting, keeping, and using data securely, efficiently, and cost-effectively. The goal of data management is to help […] optimize the use of data within the bounds of policy and regulation so that they can make decisions and take actions that maximize the benefit to the organization.

That means the goal of data management is to maximize the benefit of the organization. We defined the benefit of the organization as the increase of the voice of customer (VOC), voice of employee (VOE) and voice of business (VOB).

Thus for an operational readiness you need a high-level data management strategy which optimizes the use of data to increase the VOC, VOE and VOB. As you see, this role is a highly strategic role as it combines both business, customer and employee needs together with the data domain. Any grass root approach in trying to optimize data by data source (e.g. single software products or machines) will definitely lead towards non-optimal solutions. Data management needs to be as close as possible to the CEO, if you want to achieve operational readiness.

Data management is a truly cross-sectional function within the business as it touches all areas of the business. It touches cyber security, governance, compliance and data protection. Furthermore it needs to be aligned with the foundational readiness (data infrastructure, cloud strategy) and all internally oriented and externally oriented products and services. Internally oriented services are for example human resource management software, payroll accounting software, IT service management software systems or knowledge management software systems, while externally oriented products and services are customer service management software programs, digital or physical services or products (incl. IoT) or account management and customer relationship management.

If you look at the multitude of applications, dimensions and levels we discussed over the last few days, you might understand why it is very hard for many companies to have a decent, holistic data management. Data management is still a relatively young area for many companies and you can still find many businesses with no data management at all. You often find data silos, incompatible data formats or even no (digital) data at all. In order to tackle all of this at the same time, it is important to change our viewpoint. Instead of trying to develop a holistic data management, we should look to develop a holistic business process management, because data is created out of business processes and each process step creates specific data points. Thus if we can optimize our existing business process management, we can help to create and optimize our data management, which leads automatically to operational readiness.

In the next section we will look into enterprise resource planning (ERP) and ERP software programs which holistically try to manage and integrate a business’ financials, supply chain, operations, reporting, manufacturing, and HR activities.

Above we looked at Professor Malone’s four level of collaboration and saw that you can use A.I. as a tool, an assistant, a peer and a manager.

We assumed also that we would buy the A.I. application and run it via a cloud provider. However, when you want to make your own A.I. application, the needed level of A.I. readiness changes dramatically and the complexity multiplies.

For now we want to just focus on the operational readiness and we will assume, ceteris paribus, that you have a spirited and educated leader and a team willing to change. You have identified a great business case and know that this project will align with your strategy (Transformational readiness). Furthermore we assume that your IT team would be able to meet any challenges they face (foundational readiness).

What is Operational readiness?

If you want to make your own A.I. application, you need a small pre-project team to even assess what you need.  This team needs to be able to define the needed skills and map them to the existent and available skills in your company. How many times have we seen great organization chart with great skills added to some names and then it turned out that they already were working 100% on other projects for the next year. Thus there is always a difference between existing and available skills. Furthermore your pre-project team needs to tackle all legal and compliance issues. This also includes data protection and data security issues. Make sure to choose a pre-project team which is able to handle these legal issues professionally. What is your compliance manual saying to A.I. projects? How knowledgeable is your data protection officer? How well is your data security team able to asses potential threats caused by the A.I. project. How well are you insured towards incidents caused by the A.I.? I am just mentioning this as A.I. applications are very tricky and hard to handle by compliance and insurances. The nature of A.I. applications is that it is changing with the inputted data. Thus you cannot define a definite use-case as you might be able to do with a software application like Powerpoint. Also insurances have a hard time insuring a changing application. Before you start your first A.I. project, make sure to have talked to your existing insurance providers and check if you are breeching any defined clauses. 

Ethics

The pre-project team also has to make sure that the project you are planning is not using one-sided or biased data. All in all, all data is in some way biased, but you don’t want to have a social media negativity storm brewing over your head, just because your pre-project team was to junior to grasp the severity of the potentially tilted data.

In the end your pre-project team needs to also make sure that your company wide project management framework is set for A.I. projects. Talk to your project management office (PMO) to request changes to your upcoming A.I. projects. I have seen in many companies that the demanded processes by the PMO were hurting young A.I. projects as they were demanding information A.I projects were not able to provide. Many companies have two tracks of project management frameworks. One for software development and one for other projects. A.I. projects are however different and thus need a third project management framework (s. previous articles).

In the following section we will find out how data and data management is involved with operational readiness.

In the text above we saw that A.I. readiness is the “Extent to which a business has the ability to improve the VOC, VOE or VOB  by using A.I.”. And according to Intel you can separate it into three dimensions: transformational readiness, organizational readiness and foundational readiness. If you are the CEO of a company, you might wonder how much A.I. you want in your company. And do you want to make it yourself or buy it? What role should A.I. play in your business? Depending on that level and the decision of “make or buy”, you can determine the depth of readiness you need in your company.

Today we are going to talk about the four levels of involvement set by Professor Malone from MIT and tomorrow we will see, how “make or buy” will change the scope of your burdens of readiness.

Level 1: Tool

Just imagine you have an auto-complete function in your word document or a smart calculation function in Excel. These maybe A.I. driven capabilities implemented in your Microsoft Office and you do not have to do anything to use it. The algorithm was developed from someone else and used other data to train the model. It also “lives” in an already existing software and you do not need to build up any infrastructure on your own.

Level 2: Assistant

Facebook for example uses AI to suggest material to users to display in their feed. This again is still a supporting role A.I. is playing and is not touching the culture of the company. If you are buying this service (maybe via an API from a SAAS provider), you might use A.I. as an assistant without actually developing it yourself.

Both level 1 and level 2 A.I. applications can be introduced in the company without a global A.I. strategy as they might just improve lower level processes and a Head or Director would be more than suited to sign off on these additional features in certain software programs.

Level 3: Peers

Using A.I. applications as a peer needs involvement from the top leadership as it is the first level which deeply is integrated in complete business workflows. Even if you buy a completely ready solution, this A.I. is going to change how you are conducting business and it will change the culture of your company. Just imagine you are an insurance company and the A.I. is paying out standard insurance claims within seconds. Humans would only work on cases which are not trivial. The profile of the people having worked in this department would change dramatically as you would only need them for the complicated things in life. Many CEOs forget that some people like repetitive jobs and that they do not like everyday excitement.

Level 4: Managers

As an example, have a look at MIT’s spin off Cogito, which offers A.I. driven managerial services. While a call center agent speaks, cogito analyzes the discussion and gives advices in real time to the call-center agent – the same way a manager would do. “Speak slower”, “Let the customer talk” are just a few advices the A.I. system will give to you. The video below shows how the software works. But as you see here, A.I. truly collaborates with humans in this level and changes the culture, how people are doing business.

By loading the video, you agree to Vimeo’s privacy policy.
Learn more

Load video

Always unblock Vimeo

In the next pages we will look at how “make or buy” will change the level of needed readiness.

Everyone talks about A.I., but when you want to do it, it seems harder than you might think. Over time many people realized that a company has to fulfill certain criteria to be able to do A.I. projects. Thus the term A.I. readiness was coined. Capgemini defines A.I. Readiness as

“Extent to which a country and its institutions & businesses have the ability to reap the benefits of A.I.”

As we learned before, the benefits of A.I. are defined as improving the Voice of the Customer (Product, Services, Customer Satisfaction), the Voice of the Business (Strategy, lower cost, higher revenue, increased stock prices or profits) or the Voice of the Employee (Employee satisfaction). That means that A.I. Readiness  is the

“Extent to which a business has the ability to improve the VOC, VOE or VOB  by using A.I.”

As there are many, many definitions of A.I. readiness, I personally like Intels approach as it is relatively easy to understand and a good starting point for you to dissect the term A.I. readiness into operational, usable terms.  Intel wrote a nice birds-view whitepaper on A.I. readiness. In this paper they define A.I. readiness on three independent dimensions.

Transformational

Operational

Foundational

It is important to dissect A.I. readiness to these independent dimensions, because many companies just focus on one dimension and wonder later why their A.I. initiatives are not working.

Transformational readiness includes, according to Intel, strategic leadership, educated leadership, a clarity of potential business cases and the acceptance of your organization to evolve. Even the most spirited leader will fail, if the employees themselves boycott change. Thus a big part of A.I. readiness is change management. The further the business is originally away from A.I., the more attention we have to put into change management.

Operational readiness includes all levels of management. On a governing level it covers governance, compliance, risk but also cyber security. On an operational level it builds up human resource readiness and needs to address the issue of assessing existing skills and expertise and how to fill the existing gaps. Furthermore it includes project management readiness. This includes the adaption of existing management frameworks to yield to specific software development & A.I. issues. Additionally it includes a specific focus on data management, data flow and data resource management.

Foundational readiness is the groundwork for A.I., but should only be tackled as an output of the two above dimensions. It consists of the infrastructure platform, cloud resources, data sources and software packages. Questions in this dimension might be:

Is my existing data center set up for increased network bandwidth needed by A.I.?

Do I know enough about the offerings cloud providers have? Do I know exactly how expensive cloud services would be? Do I understand the extend of services e.g. AWS offers and how much training my employees need to work with it?

Is critical data available? Do you have it labelled correctly? Is it up to date? Can you retrieve it in real time or does your database has to calculate a SQL request for 20 minutes?

Do you have people in your organization who have worked with AI software packages? Do they know which software is out there? Have they worked with e.g. TensorFlow and know how to connect it to your existing software landscape?

So, we see the foundational readiness is a product of operational and transformational readiness. In the following pages we will dive deeper into the different levels of A.I. collaboration.