Hey folks,
I keep hearing how some AI tools are supposed to be massive time-savers, like cutting down whole workdays each week. But so far? I’ve just seen slight boosts, nothing wild. 
So, what’s the real deal? Anyone actually saving chunks of time with these tools? What are you using, and what’s it really doing for you? Would love to get some legit recommendations or just find out if they’re actually worth the hype.
Cheers for any info! 
I like Perplexity for research activities and ChatGPT for content creation. For image generation, I use Leonardo. This is the very essence of my AI toolbox.
Dorian Tireli
Poplab
From my experience it works better for text. A time saver analysing text, I give it that! Designwise it’s a disaster! if you ask for an image it creates something that looks like a stock image from '98 
“create an image about a person struggling with business admin”
good thing is, that I’ll proobably won’t loose my job as a designer just yet
I have just come around dropbox.ai for marketing - it looks awesome
I have created a CustomGPT for each of my clients so that I can get their voice and expertise when writing/thinking about stuff.
i use MacWhisper to transcribe recordings of courses/community calls so that I can then use ChatGPT to help identify engaging bits for reels or create descriptions, outlines, chapters (for youtube description)
I recently discovered OpusClip and it’s quite good, better than the AI feature of CapCut to create reels/shorts
I used Sintra bots on Make to automate tasks like replying to comments on social media (the most common one)
I use Dante to create custom chatbots that answer from the client’s knowledgebase and save their customer support team hours of work
… and many others, basically finding the best solution for the specific case
@mihai.dragomirescu - Those are some insightful tools; thank you very much for sharing!
I also use Perplexity for research and general search stuff, which is much better than others regarding the ability to backtrack its sources. It still has a propensity to make things up, i.e., give you five footnotes, and when you backtrack them, you discover that they don’t support (or even reference) the conclusion Perplexity drew. You get a handful of “pro” searches each day, allowing you to route your query through Claude (by Anthropic), enhancing the results. If you pay a Pro monthly subscription, you can do that continuously.
For GPT-type chatting and generation, I’ve found Claude to be way ahead of ChatGPT for well over a year - and recent results confirm that it has pulled ahead in raw capabilities.
I also experimented with local, offline Large Language Models (LLMs) using LM-Studio, KoboldCPP, and GPT4ALL. These require you to have some technical know-how and some mid-range to high-end hardware. The good thing is that they eliminate the data privacy/security issues by keeping all operations local so that you’re not dumping your own (or your client’s) proprietary data into another organisation’s cloud-based system.
