When will AI be able to create bespoke software for your company?


Gustaf Westerlund

June 1, 2024

“I run a consultancy i Sweden, create all the required software to run my business” – is this a plausible prompt to ask an AI in 3 years time? I think it might very well be.

Yesterday, my colleague Sara Lagerquist demoed the new feature in Power Platform where you can tell CoPilot to create an app for something, in her example it was for event management. In a few minutes with some tweeking, the app was created, with a fitting data model, UI, API and more. I think many of us saw the magic and even though there are some missing features I think this is showing us where software development is going. If you extend this a few years into the future, if even that long, CoPilot or other AI may very well be able to build software that give a decent fit to your company and by giving it feedback it can adapt the software to get a better fit.

I asked ChatGPT 4 the following prompt:

“I run a consultancy in Sweden and Germany. We are experts on Dynamics 365 CE and Power Platform. Give me a list of all the programs needed to run my business, from timereporint, managing subcontractors, invoicing, general ledger, sales etc. and a fitting prompt for each piece of software to further detail the requirements of each of these.”

It did give a quite good answer, and just imagine where it will be in a few years, and my prompt wasn’t even that advanced and connected into Power Platform.

So, where does that actually leave us as specialists on this platform? Should we all become plumbers and electricians, as those jobs are not going away anytime soon due to AI? No, I don’t think we need to. There are still a lot that will require expert knowledge that I think will take some time for AI to catch up with. Here are some and please share your own thoughts in the comments.

  1. Ownership. I think it will take a long time until we have GAI (General AI) that we allow to own things and have rights equal to humans. And I think this is a good thing. It can of course be argued that with GAI equal or surpassing our intelligence, why should they have similar rights as us? This is certainly a philosophical discussion that we need to have but I think it is far off in the future. With that I mean at least 25 years. However, AI development is “interest on interest” and hence an exponential growth, and any time estimates in these scenarios are very hard. So, being an business owner using AI will be critical, as business owners without AI will have trouble matching the productivity and hence will be out-competed.
  2. Leadership and change leadership – This is a hard one. On one hand I think leading people will most probably be better done by humans, cause people like people, in general. However, at the same time, there are so many bad managers out there and a good and competent AI might in many ways be a way better leader than at least these bad mangers. But I think we are some way off until we have AI that can show up in teams just like any other team member with a human face and interact just like a human, and lead the work that needs to be done. And as all major changes in a business requires change management, including changing any existing system, this is an area where professional services companies should focus.
  3. Mapping business to IT – Yes, AI will most probably be very good at creating the foundation of apps and also greatly increase the time it takes to create changes in the software. But there is need to really understand the business, we more often that not, challenge the customers requirements on if this really is what they need. Based on our knowledge of how businesses in general are run. Hence, the AI generating an entire ecosystem of systems needs to be questioned on how well it fits what the customer actually needs, not only what they think and say they need. This role is most often done by the Systems- or Solution Architect. One aspect of this that makes it probably a bit more important that humans do this than AI is that for some businesses, part of their edge on their customers is in how they run their business. AI will generally base any system design on how system designs generally are done, but this might be wrong as the key value is then lost. However, architects that use AI in a powerful way will most probably be way more efficient than architects that do not.