AI is becoming more and more ubiquitous in the land of Software.
Software and AI go hand in hand these days when it comes to the workflow of writing software.
However there are other important elements to cover before software is written.
Software requirements
Related tags - OpenAPI
Requirements are the cornerstone of Software, without a good understanding of the problem domain, we do not know if the software will function as supposed to for that target audience.
Requirements make up technical and functional requirements sets; technical requirements deal with performance, latency, storage and other technical requirements that can influence operation of software. Functional is asking the questions of what and how the software should work and for whom?
The paradigm we use is from Agile, there are Actors (Who) and then there are activities performed with a specified outcome (What, and Why). These “Stories” are then placed on a Kanban board from which individual tasks need to be distilled.
These Tasks are actual to-do items, e.g. Install a server, install Docker, deploy Containers with Software. These were 3 tasks that could cover 1 Story . Albeit not a very well defined one in terms of scope.
These Stories and tasks are usually iterative in nature, hence Agile. We establish a part of the problem that is known, other parts of the problem are not well understood yet. Then we first define and build what we know, and add as we’re doing these iterations.
This loop allows for customer feedback and a much regular interval, ensuring higher quality software end product than the traditional “define everything, build, and show it” method.
AI
Plays a role, for a large part, in code generation these days. But there are many other use-cases across the Software Development Life Cycle (SDLC), such as in area’s of automated testing, and deployment.
Another suite of tools that has become increasingly more ubiquitous are chat bots and how they can be used with your own data, Retrieval Augmented Generation (RAG) AI.
RAG AI enables a; “bring your own” data AI that can answer questions on information that may be proprietary, and more suitable for the task at hand.
Notable in this field is LangChain with their extensive tool-set including LangFlow. This tool-set enables quick prototyping of chat bots using a diagram like UI.
