Yes, Grok is good,
as is also building out GEMs in Google Gemini and GPTs in ChatGPT with prompts and knowledge bases.
It's all in the prompts and the personas that work it from a generic review to far deeper review.
As I tell those that I coach in this.. Shit in ---- Shit out..
The Deep research in Google Gemini for free now is amazing.. once again its how you frame the prompt before you get it to head out and pull info. good thing is it is researching with links which then you can check.
We had troubles when AI first kicked off, and I actually sent a research report I had done and posted here to the team to verify some of the information pulled. That was a few years ago, and the hallucinations were full on. Now, with the way the agents can be built, you can control this a lot more. It's black box stuff.
I am building out for tier One Construction and project management edu and licencing courses, so we are fully deep into using the tools to create useful outputs that can be verified and used in the industry.
I will be dropping these into GitHub if you're interested in playing over the coming weeks, as I have a few projects going on. The ability to program now with these tools makes it fun, but it leads to sleepless nights. While outputs can be good, there is still a lot of breaking things to get things right.
If your interested to explore more..
https://github.com/dgcruzing/Prompting-WorkflowWhile these are edu, there is the bounce to Larry's repo, which takes you deep into a rabbit hole but is well worth it once you know what you are looking for.
https://github.com/1arry1iu/archetype
Then, to build out a set of personas, this GPT of his..
https://chatgpt.com/g/g-ZR3w4e0RR-construct-cThis is my GPT in Chatgpt, which I throw things into. It is the base prompt I use across a few different LLMs.
https://chatgpt.com/g/g-Da7jidlBQ-market-maestroIf you haven't started using Google Notebook LM to load announcements and have a conversation, I suggest you do. It's also crazy good.
(20min delay)