Totally agree, got sucked back in just yesterday when I kept hoping google’s Gemini AI code would work after a few iterations
Has anyone tried using Coral ?
AI doesn’t run exclusively on electricity, it needs cash. Tons of cash. Right now Anthropic and OpenAI are making their money from subscriptions and use of their APIs. How many individuals will continue to use AI if the cost becomes significantly more than it is today?
Google Maps is currently rolling out a new version that includes Gemini on both Android and iOS. It will be interesting to see if they include intrusive advertisements.
But I don’t expect any ads in one version of this software. 
1 Like
What do you guys think about Gemini? Their new model Gemini 3 is pretty next level imo.
I’m on the pro plan and have been testing Gemini 3 for different use cases.
One of the best I’ve found so far is audio-file-to-text transcription. The results I’ve gotten to date have been excellent: it’s fast, accurate, correctly identifies multiple speakers, and doesn’t stumble when it encounters specialized vocabulary.
Nano Banana Pro (which is included in the Gemini pro plan) is also very good at creating infographics with text—i.e., it does more than generate images. Very useful if you need to craft “explainers.”
2 Likes
Looks like I need to buy a new car. 
2 Likes
I’ve been testing Gemini 3 (pro version). I like it and it is much faster than Claude, which, though I like the output, is slow as molasses. Is it just me?
That sounds cool - being able to create inforgraphics from text. I want to give that a try, but I don’t have a pro plan.
For infographics and diagrams you might also might find Napkin.ai useful. It has just moved out of beta, so it costs the usual multiple of coffees, but it does have a free level that will allow you to try.
1 Like
Just tried Napkin.ai - looks intriguing, but still kinda rough.
(Was really annoying they have an onboarding sequence that was self-defeating. I wanted to upload a file and it only gave me two choices - paste in text i copied from somewhere else (no way to type text into a box), or use an AI prompt to generate something.
I’m not exactly a newb, but had to use ChatGPT they do allow file uploading, but you have to get past the first-time onboarding and then create an empty project which then allows upload as an option.
Next roadblock - won’t accept an opml mindmap file directly, so uploaded a md dump from mindnode.
Broke it right away - did a minimal format of my text into a text editor format without any clue on the fly-outs that offered some format/generation options, but only per paragraph, not for the entire document.
Minimal help and no tutorial built- in to the product.
I did some brief experimentation and looks like it can do a lot, but this is a tool, IMHO, one has to learn and devote time to get usable output and not ai slop.
If they want casual users and not devoted business users, they will need to create a lot more self-help and tutorial guides on what it can do and how to use it.
Conversely, got so much more easily by dumping my file into Google NoteBookLM. Less options, less deeper visual models, etc. but no muss, no fuss, got a slidedeck very quickly as a video presentation.
I could be wrong, but this (Napkin.AI) seems like one of those "it’s a technology, not a product) things that will get replicated or incorporated into larger productivity tool sets.
1 Like
What a great write up - thanks. I used it in beta, so I was prepared for some rough edges, and put in quite a lot work - but did get some good looking infographics. But I think your points are valid - I hope you don’t think you wasted too much time.
I must learn how to get diagrams out of GoogleLM
NotebookLM’s “Studio” panel can produce both infographics and slide decks. You can provide specific instructions re the output by clicking on the edit pencil when you select the kind of artifact you want to produce. It’s labeled as a beta feature in the free version.
I don’t know if these are the kinds of diagrams you are looking for though.
1 Like
Thanks
I can but try
Long time lurker and only occasional poster. I’ve only just come across this thread. It’s been tremendously respectful, informative and insightful. Thank you all.
I’m also a fan of Anthropic. For anyone who’s looking at coding (from at least my very amateur level) claude-code is tremendously helpful. It iterates through
- I’ve written this as you asked
- OK that didn’t work; let me look at why
- Let’s try this [repeat]
- OK now it works.
Much much faster than I could ever do myself.
And as you can - if you feel safe about it - give it access to your local file system, it can evaluate what works and what doesn’t immediately. I’ve been using it for a few weeks and it’s already saved me massive amounts of time - or rather it’s allowed me to do things that wouldn’t otherwise have been possible. (No affiliation, revenue benefits from the recommendation etc.)
1 Like
I took the lazy way I know already with NotebookLM: I just downloaded the generated video and used frame grabs in my video editor.
For the actual project I completed, I ended up just taking the nice graphic icons and other image pieces and building the final results myself anyway.
I had switched to Claude for a bit because of this post but ended up going back to ChatGPT. Claude swung and missed hard on figuring out what an image of a badge was (trying to figure out it’s association with a military unit). And then I found ChatGPT was doing better at getting me results that I needed put in the right context.
I also prefer ChatGPTs app functionality, especially as it relates to the voice interactions.
1 Like