Found myself using gemini and the gemini apis far more the openai
ChatGPT just too slow and smaller context window
and don’t tell me that managing context better is the key - i can type 2 sentences to ChatGPT and it’s freeze for 2 mins
Thinking of a challenge: launch 1 mobile app a month (maybe 1 a week) for a year, all built in public with @v_computer.
I’ll share it all: ideas → builds → launches → revenue → expenses.
Goal? Show what’s possible with AI tools today.
Who’d be interested in following?
ChatGPTs favourite phrases:
Good catch
You’re absolutely right
Exactly, you nailed it
There’s loads more, these just seem to have increased in usage recently
Was supposed to be actually working on some database stuff
Instead spent the whole night trying to get vim-slime to auto send and run selected code to a tmux session that auto launches + attaches when I start neovim
It’s been painful, once again I should of read the docs
Anyone know when you use FaceTime it’s just quits wi-fi randomly?
I know i’m not the only one that it happens to, speaking to my friends it’s happens to them too and it’s such a pains in the ass
How is it even a thing, surely apple can fix it?
Thinking of upgrading to 17 Pro Max (still
rocking the 11 Pro Max)
If you have one, was it worth the upgrade?
Bit niche, but of the only reasons i’m actually thinking of upgrading is seeing people running some decent local LLMs at decent TPS
Full set rebuild pretty much complete
220 standing desk (160 new desk + 60 old desk)
Triple monitor (2 x 27inch 1080p, 1 x 32inch 1440p)
Desktronic Sitpro desk chair
Few more additions but just about getting there
Just spent half the morning debugging auth for an analytics CLI tool.
Turns out i’d turned on 2FA in the app at some point and completely forgotten so it’s just kept booting me
Waisted hours trying to fix something that was broke.
Trying to reduce the amount I use llms to code is actually hard
I myself just jumping to the llm as soon I realise what’s needed rather than just type it out coz I cba to type it all out
Problem is i can feel the skill draining away slowly
First visit to any e-comm site
cookie banner, discount and email pop-up..
4 clicks just to see 1 product
So much friction and just generally annoying. Brutal.
I see people quoting this tweet saying
Claude 4.5 is insanely better
I read it as, Sholto say their plan is to make it insanely better than their old model, not that it is insanely better?
big difference
I’m seeing people use LLM APIs + CLI tools to vibe code
But I don’t get it, wouldn’t the cost be insane for any decent size codebase?
You’re basically taking what was free (just writing the software yourself) and adding an arbitrary extra cost to development.
I’m quite liked the failed Ray-ban Meta demos
They feel way closer to real-world AI than those glossy 2x-speed show & tells with fine print saying pre-recorded.
Seeing Zuck & Meta just roll with it actually makes them come across more authentic.
Random thought
I wonder what training ‘methodology’ results in the biggest benefit health and well being wise as you get old (70+)
Hyrox, Calisthenics, Crossfit, Pure Strengh, bodybuilding? Mixture of these? Other?
@hubermanlab not sure if you have any knowledge on this?
Ordered a big standing desk for my new desk setup
built the desk a different room
too big to get out the door of said room door way
must now undo all the work and rebuild in office. FFS
You can always rely on Demis for a well thought through and honest answer to the question of where we’re at on the AI intelligence scale
