Please mind the gap: How to progress with AI

Over the last few weeks and months, I’ve had the pleasure of engaging with hundreds of HR professionals on the topic of AI. From these conversations, I noticed that many feel that they - or their company - are moving too slow with AI adoption.

Unfortunately, based on what I have seen over the last year, I tend to agree. There are plenty of reasons for this, but the main one is that I think we have gap:

1. Companies have too few corporate professionals or ‘domain experts’ with just enough technical know-how to design basic 'proof of concept' applications using the existing tools available.
2. Companies have too few simple tools that most people can use and experiment with.

The tools are still not simple enough

An underlying problem, in my view, is that even the simplest 'no-code' tools available in most companies — such as 微软 's PowerApps or Bubble (my personal favorite) are still just a bit too complex for the average professional user. And let's not forget that even VLOOKUP or XLOOKUP in Excel may be a step too far for many. A while ago I made a simple video on how to create a simple flow, but I'm willing to bet that most people have not tried it, or did try but did not persist and succeed. (please do prove me wrong!).

The number of people who persist and learn how to effectively use these tools, I think, is still too small. For instance, I once tried to create a 'simple' Microsoft Flow to populate a Word document, but it wasn’t as easy as I initially thought. However - if you know me - that only made me more persistent to get it working. I do admit that in some situations I would just give up, like the time I tried to fix my dishwasher and accidentally short-circuited the electronics. Purely coincidentally, I decided at that very same moment it was time to buy a new Wi-Fi-enabled dishwasher anyway. So in the end it was a win-win, I got a new dishwasher (self-funded - which I prefer to ignore) and I obtained valuable new skill of knowing how NOT to touch certain sensitive electronics.

So start but don't stop

Anyway, coming back to my frustrating evening: it took multiple attempts, reading, and plenty of failures before creating that Flow became "simple" to me. Even platforms like Microsoft PowerApps or Co-pilot Studio, while powerful, still require just enough technical knowledge to intimidate most users. But, we really should not expect corporate professionals to understand how APIs work, Function Calls, JSON structured output, Python, how ' entities', 'nodes' or 'topics' are use in Copilot, FormatDateTimeStamps formulas, when they best can use fine-tuning, large context or RAG. That is just too complicated for most people.

Mixed teams may work but things can get lost in translation

Then, an often tried alternative is create teams of (paired) domain experts and technical builders that work together but – in my experience - that means things can get delayed or ‘lost in translation'. Or it leads to creating a project team of a dozen people with more time being spent on project management then actually building the tool. Now, if both professional experts would know just a little a bit more about the 'other side' however that would become so much easier but that's not that simple; as I explained above. It requires the non-tech person to have a bit of tech knowledge and the tech person to have sufficient domain understanding. Plus, both could use a good dose of inquiry skills - since both dont know what they dont know.

Hence, a better and more efficient solution, I think, is to make no-code tools simpler and more widely available —allowing corporate professionals to develop basic AI solutions - within a safe environment, while still offering them an option to make things more complex for those who want to dive deeper.

Paper and crayons

To put this in perspective, imagine you’re a born architect. You have a clear vision of a perfect house in your mind, but explaining that vision to a client is tricky. You might say "open kitchen with a wooden floor" but what you’re imagining is different from what they are. You end up explaining and discussion it for four hours but still they seem to understand it differently. So you want to visualize it for them. Now, suppose you’re (only) provided a top-tier computer with the best CAD software and an ergonomic 3D mouse.

In theory you could design anything you want. But, in practice, you don’t know how to use CAD, and navigating hundreds of menu options feels overwhelming. Most people would probably just give up. A small percentage (1%?) would persist, master the tools (probably after a year of frustration), and produce great designs. But what about the other 99%? They might have had even better ideas, but without the right tools, their vision never materializes.

What if, instead, you gave those 99% a simpler tool — a tablet with a pen. or even just paper and crayons? Their creativity could shine!

And importantly, let's not forget that both the 1% and the 99% would not build the house themselves — they would only hand their ideas to specialists who could translate those ideas into building plans and, eventually, have the builders do the actual construction. So don’t expect architects to be builders, nor builders to be architects. Instead, equip them with the right tools to communicate and collaborate effectively and bring the vision to life.

Don’t expect architects to be builders, nor builders to be architects. Instead, equip them with the right tools to collaborate effectively and bring the vision to life.

In sum, my recommendation: Empower your workforce with easy-to-use, safe tools that allow them to build and experiment with AI solutions. Expect and celebrate failure. And, for those (few) successful experiments that turn out to be ‘good enough’, have those handed over to larger mixed teams (with the initiator in the driving seat) and make them robust, safe and production ready.

Now, what about that 1%? Well, the good news is that you already have them, I'm confident of that. You just don't know it, and the 1% may not even know it either. They could be that new joiner, or that senior executive, or that person working in a remote location. So you only need to find them, give them the tools and then remove as many barriers for them as possible.

Then, in both scenarios, you may end up with a beautiful vision brought to life. If that is what you want.

Note: The statements, views, or opinions expressed in this article represent my own views.