AI Strategy: An introduction by Steven Nooijen

What Are the Differences Between AI & Data Strategy?

A data strategy and an AI strategy are compatible. AI strategy is a bit more offensive, so you look for applications that can generate value, for example more sales or cost reduction, but it generates money. Data governance is a bit more defensive. So you try to prevent a data breach or a data leak in some other way, or that people get access to data that they should not have.

Defensive is difficult to quantify and build a business case for. That is why we often say start with an offensive use case, i.e. the AI ​​strategy, and follow your data strategy from that to build up your defense properly. If you have already run a use case and you improve the data quality, you can immediately measure the impact of the data quality in the use case. This makes defense directly measurable and quantifiable.

Several studies actually show that if you ask data professionals or business people in the field, 90% say that data is important and has potential value, or is a business opportunity for the company. At the same time, only 15-30% say they are really getting the value out of data that they think is in it. So there’s a huge mismatch between what’s in it and what’s actually being achieved. That is exactly why you should deploy a data strategy.

In recent years that we have been working with data, we have all seen that you’re not there yet if you only put data into a data lake and hire a data scientist. The added value of an AI use case only starts to come in when the insights from the data are processed in the workplace. If it is adopted in the business. Precisely that bridge between what happens technically and is possible and how it works in practice and the implementation in the business, that must come together and a data strategy can help a lot.

We believe an AI strategy starts with your business objectives. Where do you want to go. Certain use cases emerge from this, certain initiatives that support the business strategy. Those use cases are made explicit in your AI strategy. That results in kind of a product roadmap. If you have those products clear, you can translate that into what you need in the sense of your capability. So you have people, data, tools and techniques. You can also determine strategy about that.

For example, are you going to hire people yourself or do you work together with a consultancy? Do you buy certain AI ready-made applications or do you build it yourself? Do you organize your data science centrally or decentrally? Which cloud vendor are you going for, or are you still doing it on-premises? What kind of architecture do you need? These are all choices that are ultimately subordinate to which products you want to do and which business objectives you want to achieve.

At GoDataDriven we have our definition of gone. So when we leave, we always know for sure that we can leave.

That’s because we’ve neatly handed it over to someone else or delivered a piece of work that’s just finished. Our experience is that customers are always very happy that they can continue independently and then thank you very much at the end of the process. Of course, a satisfied customer that can proceed independently from third parties is why we do it.

The Role of an AI Strategist

Steven is a strategy consultant at GoDataDriven and has been working with us for almost five years now. Before that, he had been working as a data scientist for another three years. After five / six years of being behind the keyboard, he has progressed more and more into a management consultant role in which he leads and coaches data teams. Last year, we introduced a strategy branch within the company, which is Steven’s role to set up.

AI strategy involves making plans to achieve business goals with data and AI. The roll touches on technique but is not technical. I am no longer in a development rol. Sometimes I think that’s a shame, but it’s also very nice.

It is much more about telling stories, taking the customer along and conveying a message. The technology has become Powerpoint. The Powerpoint is about code, data, technology and architecture. It is also important that you have practical experience in that field to fulfill this role. Over the years we have seen that you do not get there with just people, data and technology, there is also a bit of organization involved.

Precisely because artificial intelligence has a lot of consequences for business and business processes, you see that the workplace must also change. The business must take the insights from data into account. That is an entire change process, it is much more about change management than just implementing tooling and technology. That makes it difficult.

We believe there is no standard path to the role of an AI Strategist at GoDataDriven. Whether you come from a client, or another consultancy, or a start-up, it can all lead to the role of an AI strategist. The only important common denominator is that you have experience with data, that you know in practice what works and what doesn’t and that you can convey that to your customer in ‘Jip and Janneke’ language.

In an AI strategist we look for curious people, people who have a great passion for learning and who find the field very interesting. But we are also really looking for people who are talkers, have strong communication skills, seek connection with others, and who enjoy taking others along in a change process.

It’s not always easy, of course, so you’ll also have to be straight forward now and then and give feedback to the customer in more difficult times. Ultimately, it is people’s work with a focus on technology.

What Does an AI Strategist Do?

We want to go beyond making a stand-alone plan. We want a plan to be very concrete, actionable and pragmatic. And then we also want to put our money where our mouth is, by helping the client after we have developed the strategy. We do this in the form of coaching programs in which, for example, we work with the client one day a week to monitor the implementation of that strategy or to implement specific technical use case.

The nice thing about it is that it is not just code, programming, and implementing tooling, it is much more about how you ensure that value is actually obtained from that piece of code and application. The perspective needs to be broader.

You have to look and say “OK, how are we going to ensure that that application will soon be used by people on the work floor”, and then include the people in the process.

A certain role may be that you are a reviewer, so you are asked to assess an organization’s current performance. In a very short and intense period, you will determine how that organization is doing when it comes to people, technology, data and processes. Based on the findings, you can advise how that organization could improve on these areas. That is the role of reviewer. If you can then help a customer implement your action plan or steps, you will become a bit more of a coach over time. If you really work together on a strategy for longer, we are more of a partner. Where we really spar with the customer at a high level about how we are going to make data successful within the company.

What appeals to me most is the trust you get from clients when you can really show that you bring knowledge of the field and can apply it to the client’s domain. Being a sparring partner appeals to me. What really helps in this is that helicopter view of how how a particular client is performing and how the field is moving. Bringing all this together in one case is fantastic.

Many organizations still do not know very well what data and AI can and cannot do and what it means. Even Steven thinks he did not knew it that well three years ago. He now has a lot more experience, and it just takes a lot of time and experience to understand what makes a project successful and what doesn’t. There are certainly a lot of myths involved. What it mainly comes down to is managing expectations very well.

Hiring one data scientist straight out of university, gives little perspective on a successful implementation of data science for an organization. A larger playing field of people with different skills is required to do that. Getting that message across, that is my daily work.

The biggest misconception we’ve come across about AI is that an organization thinks it’s more of a project than a transformation.

“As long as we hire an external vendor, we can buy AI and it will work itself out. We don’t have to worry about that anymore,” they think.

Precisely because of the nature of data that changes continuously, your artificial intelligence applications will continuously change and it is therefore a lasting investment. That’s really the biggest misconception we regularly deal with.