Investigate differences in prompt formation between GPT and Claude.

[iramykytyn]

Read prompt engineering approaches specification overview for Claude and GPT:

• https://docs.anthropic.com/en/docs/build-with-claude/prompt-engineering/overview
• https://platform.openai.com/docs/guides/prompt-engineering

Try different techniques for our prompt for AutoScan project and document the results. Reference this in your previous article for #4 .

[eLQeR]

All the differences that were researched are described in the Google Docs below: Differences in Prompting Techniques: Claude vs. GPT

[iramykytyn]

Reviewed the article. In general good job, but few things I would add:

• make the article more specific and show how each approach affected our tests result
• keep reference to sources where some of statements where validated (like other articles, LLM's documentation, etc).

Regarding the first option, I'm not sure it is possible for the current size of the data set to demo the difference taking into into account the randomness we observer, but at least it would help to show the result when we have one generic prompt vs result for separate prompts for each LLM.