insights

How to get started with AI: The six step framework

Jochen Derwae

Ever since OpenAI introduced Dall-E back in 2021 and especially ChatGPT in 2022 there's been an - almost - explosive growth in interest for AI. The generative AI revolution has opened up the application of AI in many more tasks and industries than ever before.

Before this decade, there were already some very impressive results when using AI. Whether you were predicting sales figures months or a few years in advance, trying to detect goods with production errors on a conveyor belt or suggesting movies on a streaming website, AI had already found its place when working with numbers or images.

Now, this fantastic technology has also found its way into many text processing tasks or even generating entirely new text or image content. And the number of potential business applications is growing by the minute.

With the new generative AI you can now easily generate images or videos based on a description, answer emails automatically while looking up answers in your company’s documentation or getting inspiration to write the next report or vision document.

“Ok, I get it, AI is powerful and delivers great business value”, I hear you saying, “but how do I get started?”. That indeed is a very good question. Here at SUMSUM we’ve been helping companies with answering this question for years. In fact, we’ve developed a methodology to help you along. This series of 6 articles will explain this methodology and give you some basic tools so you can get started on your path towards a great AI future for your company.

Introducing the 6 step framework

We use our in-house developed 6 step framework to coach companies from the first tentative glances at AI to successful AI product roll-outs. Here are the six steps:

  1. Get ready
  2. Ideate
  3. Assess
  4. Identify
  5. (Build a) Pilot
  6. Iterate

Get ready

Immediately starting to build an AI application in your company without being well prepared can have disastrous results. As Gartner stated several times before, 85% of AI projects fail to produce a return. On top of that, organizations that had to deal with such a “failed” AI project often conclude that AI is not that useful or not for them. And that really is a shame! They are missing out.

So, how do you prevent your AI project from ending up in that 85%? Well, it starts with good preparation. Here are several things you can do:

  • Learn more about AI
  • Investigate whether your company is AI-ready
  • Train your people (or hire specialists)
  • Find a partner

Learn more about AI

In order to drive a car, you don’t need to know how a combustion (or electrical) engine works, what electronic circuits are in it or how gears are created. What you do need to know is what fuels your vehicle, where you can and can not drive and when to go to the garage for maintenance.

The same is true when you’re at the driver seat of a company, you don’t need to know the ins and outs of how AI works. You do need to know what can be expected of an AI tool (and what not to expect), that you need data (and often lots of it) and what the legal and ethical limitations are to using AI.

Here are a few ways to get started:

  • Online AI courses and articles
  • An inspirational session by our friends at Raccoons
  • Watching some youtube videos - I can highly recommend the videos by Cassie Kozyrkov who has great organizational level explanations and tips relating to AI.
  • Get some training at a trade organization or university

Investigate whether your company is AI-ready

An organization needs to be at a point where they can support the transition of (some of) their internal processes to AI driven technology. More specifically you have to be sufficiently advanced in all of these aspects:

  • Strategy - how do strategic plans get developed and distributed throughout the organization, have you defined KPIs and are they monitored? Can you adapt your strategy flexibly based on new numbers or predictions?
  • Data - is data stored and readily accessible for further processing, are the relationships between different data sets defined? Is data managed centrally or by the departments?
  • Technology - are you sufficiently up to date with your software and hardware? Can AI applications run on your systems, is a cloud connection possible?
  • People - do you have people who have analytics or AI knowledge, are they in one centralized group or do you have AI champions dispersed across the different departments? Do your employees have sufficient freedom to experiment with AI solutions and to identify those processes that can be improved?
  • Governance - do you have systems in place to automate the release of (AI) software, can you measure the performance of a new AI system?

In order to help with this investigation, we at SUMSUM have developed an AI readiness test AKA AI maturity assessment. Discussing this in more detail is a topic for a whole other series of blog posts, something for a later time.

Train your people (or hire specialists)

If you want to make AI projects a success, you’re going to need people who at least know how to get the most out of an AI system in their day to day work. There are several ways you can organize this: have a centralized AI team within your IT department, have several champion users dispersed throughout the different departments or a combination of both. In any case, you’ll need people with enough specialized skills to accomplish this. You can do this either by hiring AI specialists or by training your own people.

It’s our advice to have a combination of a centralized team that can set up a way of working to release and monitor analytics and AI systems and provide specialty help with specific questions but, on the other hand, have data scientists embedded in your departments, supported by a department staff that is knowledgeable about AI.

The advantage of this approach is that the data science knowledge and the domain expertise are working closely together. Also, there’s no centralized department that is the bottleneck to delivering business value. An embedded data scientist (or team) can respond in a more agile way to the department needs.

Find a partner

Finally, find a partner that can help you with your AI projects. Whether it is to do (partial) staffing, provide specialty expertise or get you started. Look for a partner that can support you with the whole software development lifecycle and that has a large network of specialists.

Obviously, I’d like you to consider SUMSUM as your partner in your AI endeavors.

Take a look at step 2: Ideating

more insights

Close Cookie Preference Manager
Cookie Settings
By clicking “Accept All Cookies”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage and assist in our marketing efforts. More info
Strictly Necessary (Always Active)
Cookies required to enable basic website functionality.
Made by Flinch 77
Oops! Something went wrong while submitting the form.