The LLMs are multiplier of efforts, cogs and pulleys, not oracles. They are bicycles that allow experts of a context to do 10x the amount of work we used to do.
About the conversation around automatic programming, how useful or useless it is. Well, for me it is quite simple: you need to be very good to multiply your output. Retaining quality is harder now, but if you manage you are accelerated 100x. Instead of fighting I enjoy the race.
@thsottiaux Hi! Performance are ok on my MacBook. Can I request to be able to put side by side the chats using drag and drop like you do on VSCode editor. I want to see multiple unrelated chat at the same time.
@dteam69@shrav_10@adxtyahq Up! I don’t see anyone complaining about the re-sending explain how it should work in the opinion as the LLM are stateless, except the weights during the training. They would like they cache the state of the llm after the inference and you provide just the n+1 message, I assume.
@theo Said that, it sucks, tbh. I hope they really change direction because they are enshitificating the product because subsidiaries and incentives are hitting hard.
@theo The answers you’re getting look like our internal chatbot with our custom system prompt which is meant to direct users to our internal (non-tech) procedures and knowledge. I suspect they changed recently the system prompts/prompts to make it more “coding-directed” than Mac use.
@macastel3 Not just that section but also the normalisation of the robots in the home of the people and, even, robo-actors in the comedy scenes.
Crazy how fast, in China, they are make the future real, without cry every days about “the future is killing us” typical of other countries.
@theo Think about that now is even better then 5/10/10+ years ago, imho. I literally cried for one week to put my first app out when I put my first app on AppStore at that time.
@macastel3 Got a cup earlier, actually.
I find funny that the price is never the one they display but always lower and discounted by 50% at least, no matter what you get and when. They always want to give you a paper bag even when you get just one cup.
@macastel3 Point is how much is the same flat in World tier1 town central location? E. G. London. Moreover, how are the conditions of these houses? For what is my experience China wins for this “premium” location - at least these are renovated or made recently. While in London… 😆
@linuz90 I really love to notice how every agent has kinda a different “personality” in reviewing the code which is then a quality in choosing the possible inference path which is probably because the prompt that the main agent give them.
@linuz90 Many times I want to work on backend and frontend at the same time and they are in the current monorepo or I have a flow of ideas and I want the typing bar free and available.
@theo Congrats! I basically discovered you during 2025 and your videos have been so useful to learn how to integrate LLMs in my company workflow. A big thanks to you and T3chat to inspire me along the year to make a better chat UI than the big guys ones!
@steipete Yes. That’s some of the reasons why I asked before spending time on it. According my experience it would map nicely, but I’m not super expert of GitHub.
I’ll have a look if I’m able to squeeze out some spare time. BTW I like how it’s coming along!