There’s a cottage industry of AI gurus who extol the virtues of being open about the advent of AI in the enterprise. Don’t be shy about it, AI is powerful enough as it is, and it can already help in the current state of its evolution (mainly, powerful chatbots that have access to certain “tools”, and vendor APIs with which to build some more customized applications, backed by cloud LLMs).
It’s true that AI helps, and you’re probably already extracting some juicy bits of productivity with a chatbot, whether it is officially allowed or not in your corporate context.
But the moment what you have in mind becomes more ambitious, or more specific than that, in terms of the exact problem you are looking to solve, you are likely entering uncharted territory.
There is something called “business rules” (or “business logic”), that normally comprise the set of procedural, written and unwritten, legal, technical and all sorts of other adjectives ending in “al”, that come with any process of any complexity, operated by a group of people, in any corporate or institutional environment. And more often than not, applications created to operate in such environments are made to “encode” these rules (either explicitly or implicitly) or at least espouse them. But if you think that the current AIs are simply powerful enough to understand, by themselves, all the rules in your context, that you can simply throw a bunch of PDF and PowerPoint documents at them with the hope that they will figure it out, you are deeply mistaken. They will figure something out that’s for sure, but it’s very unlikely that it is what you really have in mind.
Building an application with modern AI does not dispense you from sitting with people, to establish what exactly is the problem to solve (very often much less trivial than what is assumed) and how the components are going to be defined and orchestrated, in order to be able to make use of the AI in a more realistic way than the “mind reading magix box” paradigm than is being sold everywhere with too much enthusiasm. You still need to think and care about business logic, there is no escape.