How to successfully launch AI projects in your organization

I was recently asked to speak at a conference on this topic: How to successfully launch AI projects in your organization. After giving some thought and structure to my speech, I thought it might be a good idea to share some of my thoughts here on LinkedIn.

1) The What

It seems strange, but one of the biggest changes so far is getting a proper definition of what you want to do. The worst thing that can happen to you is that some AI "expert" consultant tweeted some ideas into your CEO's head and now you have to deal with it.

If you want your organization to be ready for AI, you don't need coders who know Numpy, Tensorflow & Co. You need people with a broad mindset who understand what can be done with what effort. You need people who understand your business, its processes, and its challenges, and who also have an overview of what is possible with today's technology. Once you have these people as part of your organization, you have a good starting point for what you can do.

Forget all the companies and consultants trying to sell you AI solutions. Most of the time they are not technologically mature enough and, most importantly, they do not understand your business.

2) The How

Today, the answer to everything in the IT world is Agile. That Agile is not the answer to everything has been written by me in some articles in previous LinkedIn articles. But especially in AI development you need a much clearer framework. You need to define your deliverables up front, and you need some classic waterfall timelines to work on as a roadmap. Otherwise, your projects run the risk of getting stuck in a development cycle forever. In particular, training, annotation, and model refinement are things you can do forever.

Keeping your models open and working with a good git will ensure that you can follow up on good ideas after you have delivered your first successful AI applications, but do not deviate too much from your initial deliverables. Especially when working with machine learning algorithms, it's a good idea to use reinforced models rather than one-off annotations. This allows your application to stay current and responsive to changing requirements.

3) The Who

As already mentioned above in the "What" section, having your own people capable of understanding your business processes and the technology is key. They are hard to find in the market, especially today where everyone claims to be an AI and process expert. You as the CIO/CTO are responsible for having the right people in place. So my advice would be to first look for people who have the right mindset to fit into your business culture. They can be junior or senior professionals who are able to understand your business, and they should have a good understanding of technology. And not necessarily AI.

These people can be trained on ML or Generative AI. You have to make sure that you raise their imagination and that they learn to connect your business problems.

Once you are prepared on these first 3 steps, I guarantee you that your business will take off successfully in the AI world.