I’d like to go inside the AI contexto, but I’m newer about it.
Context:
1- I have a list of valid domains for emails (gmail, hotmail…)
2- I have a list of invalid domians confirmed from ESPs (not exist)
3- I’ll have every day a list of new domains to check.
If match with set 1 or 2, the domain will be removed from table (join context) - were validated
With the table result, I’d like to use the AI to watch from invalids domain to finde something “seams / Like” a invalid domaid to mark it to validadtion.
@denisfi here is an example how to use (local) LLMs to do chunks of tasks and collect the results in JSON structures to (re-)use them later. Maybe you can adapt that approach.
The functions depend very much on the sort of LLM and the prompt. One idea can be to let AI write the prompt.
Hi @mlauber71, thanks for your quick reply and guidance on using an AI. I’ve never used one, but I think your suggestion will be very beneficial to the case as an alternative to identifying case variations from the real data.
When it comes to domains, you have seen that there is a very wide range of slight changes that can be a lot of work to analyze and validate manually, and having this guaranteed, it is easier to separate the correct ones first and then test variations of them and use the invalid ones as well.
I’ll go into more detail about AI later, as I know it can help with much more than command prompts.
If you have any more material on AI components and features that you can share here, I know it will be very useful for our fellow forum members.