ИИ в работе системного аналитика: от промтов до ТЗ

Практический разбор на примерах: сбор требований, диаграммы, Use Cases и ТЗ. Плюсы, минусы,подводные камни и промт-чеклист для системного аналитика.

https://habr.com/ru/companies/auchan/articles/1009920/

#ИИ #системный_анализ #системный_аналитик #промтинжиниринг #нейросети #chatgpt #искусственный_интеллект #plantuml #use_cases #сбор_требований

If anything at all, this #AI thing is bringing to people attention the existence of some #plaintext esoteric syntaxes, such as #markdown, #plantuml and #mermaid, among others I myself find funny or exotic.

Having an average knowledge of these formats make interactions more straightforward requiring less "usage", or however the interaction scarcity is called.

Also, the effective use of plaintext in prompts, in shared attachments and in required outputs (instead of overcomplicated proprietary formats such as #microslop office formats), greatly increases the response speed and efficiency of #LLM s.

I don't know if anyone else (still) uses PUML, but I'm a huge advocate. It's like the markdown of mind mapping.

#PUML #PlantUML

On a side note: before you spend an hour debugging why #PlantUML does produce colours, make sure you didn't include a colour theme that's black and green.

Just saying.

So I spent some time in #PlantUML trying to model #Risk #Trust #Attacks to tie #TOGAF #SABSA #Threatmodel and #PASTA #STRIDE together

As a Hacker, I don't need a fancy #diagram to visualise the attack vector and path I am already seeing. But as security architect/Team Blue I need to show it to others who don't see it.
And it has to be audit compliant.

The public repo with my PUMLs is here, feel free to reuse it
https://codeberg.org/0xKaishakunin/Architecture/src/branch/main/PlantUML-ModelingTrust