Member-only story
ChatGPT and Software Architecture
Let’s generate diagrams
There has been a lot of interest in ChatGPT from OpenAI in recent weeks, with all sorts of interesting use cases emerging.
In many ways, this is the AI equivalent of the architects whiteboard but has numerous uses than just drawing lines and boxes. With this article, I want to use this innovative technology to demonstrate another use case for starting the software architecture process.
Just like drawing on a whiteboard, the process is a tad chaotic because of the typical interplay that causes constant revision to find the best answers. This article aims to demonstrate some questioning techniques I've used successfully with ChatGPT that have produced results of genuine value.
One caveat with ChatGPT is that it is still learning, so don't expect consistency. Asking the same question on another day may yield a different result.
Setting up a ChatGPT account is simplicity itself, which can be done by following the instructions at this address: https://openai.com/blog/chatgpt/
User Story
So let us start by asking ChatGPT to respond to a user story.
write a user story where a surveyor logins into a system and enters information about the geospatial coordinates, the building, it's type and it's characteristics of just one of the many surveys they will perform in a dayAs you can see, this user story is specific to my industry. ChatGPT has a vast knowledge base, so experimentation is required to understand ChatGPT's depth of knowledge — what does it know?
The key here with the initial question is to be very explicit about crucial details to get the best results. This way, ChatGPT can respond with a decent set of results such as this:
If your results are anything like the above, then you know that ChatGPT has picked up the trail and will build on further questions you ask. The key to remember is that all the following results will be based on this initial context, hence why it is worth spending time defining the…
