AI can do amazing things. If you did your homework

It is so impressing to see what AI can do! Do love to watch demos and see how slides are generated like nothing, pictures appear, all these wonderful things? I absolutely do.

And then, one goes back to real life work, and, well, every day life is like every day life. Things take time and effort. In contrast to product demos.

Actually, when I do a demo, everything is scripted to the extent that I have usually have a backup video. Just in case.?Because these things work. But not all the time and not with every dataset. And specifically not if you have no backup and important people are there to see it. Right?

In real life, there is a lot of work to be done before the data starts to dance. And, while it is always OK to dream a little, at least the tech folks know this by heart. There is always some pain before the fun starts.

But: what exactly? And how can Business and IT cooperate best to get it done together?

?Well, at DATAHEARTS, we believe that it always starts with a strategy, followed by a merciless assessment and then good software. Not rocket science, but three simple steps:

?? Where do I want to go to?

? Where do I stand?

? What's the best way to get there??

?

Where do I want to go?

The first, and maybe most important question is the purpose question. Most companies have a strategy deck available, and that is a very good start. So, the first thing is to sit down and reflect how the project at hand fits into the overall strategy, and hence what the purpose is. Enabler? Customer service? Cost savings? Sustainability?

It has to be clear how this effort supports the big mission. Take the time and clarify questions with senior management or responsible teams. Only if the purpose is clear, the team can make good, fast decisions in difficult moments. And it helps a lot to keep motivation high.?

Let’s assume that your company strategy is to focus on customer satisfaction. Now, one underlying goal is to reduce the time that a customer has to wait for an answer via email. Obviously, one idea could be to generate send-out-ready emails for all standard requests, so that the agents only need to crosscheck and send, instead of writing the email.?

Where do I stand?

Now, the goal is clear, makes sense, fits the strategy. So, we need to find out what we have, and what we needed to make it happen.

There will be some tactical requirements, such as templates and access to a price list for cost related questions, the company adress for others and some kind of a chatbot. But, if you just grab those and start to create a solution, it will not deliver at quality. There will be too many mistakes like not implemented restrictions and exceptions, maybe security risks, or generating the emails will take longer than just writing them? Anyways, likely the team will eventually start to delete all prepared offers and go back to the old process. Because it may have been slow, but at least it works.

The better way is a comprehensive assessment. For example the framework of DAMA DMBoK that our partner Connected Data Group, een Open Line bedrijf . I feel it helps me as a business/IT hybrid to understand where we stand.

At DATAHEARTS , we recommend to work with comprehensive frameworks to make sure we at least consider to evaluate each element of the data foundation. We may then skip something consciously, but we don’t forget it.

To stick to our example, we may have to check if our architecture allows to combine client names, their history and other metadata with information from the internet, such as market developments or the latest production cost which might be relevant given the energy cost. We may have to work with the purchasing department to find out what sources they access (incl phone calls, website visits and little, hidden notebooks) to learn how they actually find out to research the content of their emails.

How do I get there?

Once the complexity is understood, execution can start. Business may start to evaluate use cases, align the team and guide the change process.

For data, the team can now use a structured approach to close all gaps identified in the second phase. If, for example, it turned out that security, overall architecture and modelling need some care, but the rest is in shape, this can be visualized well using the wheel shown on the picture.?

For business people such as myself this is so helpful. I can see that the team follows a structure, some items are covered, the rest is moving along. Even if I don't understand the details, I can still follow the high level thinking, and understand where we are in the process.

Example: If the data project lead tells me: “Look, we thought through all 10 elements. Overall, we are in good shape. But: we should do a deepedive for data quality, security and how we model it.” I understand where they are. Seven out of 10 done, that's good! And detail work in some areas. Got it.

In contrast, the standard answer: “OK, thank you, we got your requirements, let us check internally and come back to you.” leaves me nervous. And makes me sit down beside you to follow up every 5 to 10 minutes max. Yes, I can be a pain...

Summary

In summary, I can only recommend to select a good framework. As we work things out, this is so helpful to keep everyone aligned and high level informed about the progress.

I really liked this one when Erik suggested it. I think it covers all major areas in a good way and helps to streamline the conversation. But there are sure so many more.

What's your experience, what do you use?