Microsoft Power Apps could be improved because there are still a lot of jargons and too many moving parts. For example, if you look at Copilot, the term Copilot is confusing in the sense of whether it is Copilot in M365, Copilot Studio, or Copilot in Microsoft Power Apps. There is a plan designer which uses Copilot. The whole thing how AI has been positioned is still not lucid for the end user. An end user wants to know exactly what they want and where they go to get it. I think that could also be because things are evolving so fast. From an end-user perspective, the way it has been positioned, the clarity and the boundaries between the different types of offerings and AI offerings available is confusing as of now. There should be better clarity on that. The biggest issue I have, and I have also spoken to a few of my clients about this, is the licensing model. In traditional software development, almost 95 percent of the time, the development team bears the cost of the licenses. For example, if I develop something, I may have to pay licenses for four or five different software that I use. As a user, if you use my services, you probably pay something to me as a subscription, but you do not have to bother about the licenses. All that is wrapped under the hood. Unfortunately, in Power Platform as such, and even in other low-code things like UiPath, if you use a premium feature such as Dataverse, almost everything ends up using Dataverse or SQL Server or some relational database. If you use that, then as an app builder or app maker you have to have a premium license. The end user too would need to have a premium license. That really makes the adoption prohibitive. It is too expensive. We are talking about something like around just for Microsoft Power Apps alone, approximately twenty dollars per month, which is extremely high. Another point to consider for what else can be improved in Microsoft Power Apps is that one does not know what compute power one is getting when one buys a license. If you look at the licensing model, you will get to know how much of Dataverse storage you will get in terms of log storage, database storage, and file storage. However, you do not get to know how much of compute power is being given to you. I do not think Microsoft has an SLA saying that any request of a certain amount, such as MB per second, you will get a response time of whatever, one by sixtieth of a second or some millisecond. I do not think that they have that performance SLA in place. They do have storage SLA which comes with the license, but they do not have a corresponding SLA for performance.
