Soleone

[Soleone @soleone]

@Dogma87x @Bimon Mit der 20 EUR Subscription für Anthropic (Stand heute etwas besser) oder OpenAI kannst du schon einiges machen, vor allem wenn du regelmäßig neue Sessions startest damit das Kontext Window nicht zu groß wird. Kann dir gerne helfen wenn du mich anschreibst.

[Soleone @soleone]

@strimblez Clear language, frontend design, creative riffing, and just pushing through to the finish line.

[Soleone @soleone]

@udaysy @OpenAI @OpenAIDevs I don't know, but sounds related to this? https://t.co/tsvUl0lRQA

[Tibo @thsottiaux]

There were reports of usage draining faster than anticipated for some users over the weekend. We've identified approximately 40,000 accounts that had an inconsistency on which usage limits where the actual usage that was applied did not correspond to their actual selected plan. We are reaching out individually to all these users and have corrected the issue. If this affected you, we'll add credits to your account. Sorry we got this wrong.

[Soleone @soleone]

@doodlestein I am just trying to give an opinion from an end user, in an attempt to help potentially improve the popularity of your products. Was well meant. It's all good!

[Soleone @soleone]

And then using something like https://t.co/fzGdjx9WSK to quickly pick or discuss individual suggestions.

[Soleone @soleone]

An AI incantation that has surprised me producing good features in a short amount of time, especially with Opus 5.5, is saying "riff with me on xyz".

[Soleone @soleone]

@DCinvestor I get a lot of stuff done using the $20 sub. First I only had OpenAI, now also Claude for the amazing Opus 5.5 design skills and clear language. It lasts a good amount when optimizing the input token cache and context windows, e.g. with custom compact commands (like using Haiku).

[Soleone @soleone]

@DaFkYouMean @MyCrazyIdeal @TngPointless @thsottiaux If you "ping" it, that means sending a new tiny prompt which still incurs the cost of all the cached input tokens every single time. There was no way to bump the cache expiry for free.

This genuinely seems like a big development and hopefully something that catches on.

[Ben Davis @davis7]

I've started to think of design as a proxy for how good a model is at making decisions that can't be easily represented with a verifier function

These decisions are the "taste" ones, the "creativity", the "cohesion", the "magic" that Claude models are leaps and bounds better than anything else at

For most of this year the "claudeisms" have been insufferable because the models weren't good enough to take advantage of the personification they'd been given.

I think at the previous level of capability OpenAI had a huge advantage because of their model's super autistic nature they wouldn't fly off the rails doing weird shit and weren't super annoying to talk to

This flipped with Opus 5.5 when I think the Anthropic way of making Claude really paid off as the model got way more capable and could use it's more "personified" nature to make better decisions without a lot of the previous issues.

Astra had a lot of super weird behavioral issues. It's capability is incredible, but it's super literal nature makes it so much harder to work with especially for the kinds of long running tasks we're all doing now. It makes worse decisions than Claude and at this new level of capability that's starting to matter a ton

OpenAI has a lot that needs to change and be fixed at a pretty fundamental level imo to get back in 1st. I think they can do it, but it's gonna need to be an Astra 6.5 level jump that really, really changes things

[Soleone @soleone]

How to build great products using AI:

1. Create a basic prototype.
2. Use your product relentlessly.
3. Immediately create a note when something doesn't feel perfect.
4. Work off all notes.
5. Go back to 2 (which is the crucial step).

Do this for weeks until it's polished.

[Obie Fernandez @obie]

The codebase is the prompt

[Soleone @soleone]

@PovilasKorop To be fair, they are different from prompts in 2 ways:
They are using progressive disclosure to decide if they should be expanded and then if they should be further expanded into potentially many sub-files. And they can also ship with scripts. Personally I'm fine with the term.

[Soleone @soleone]

logdig 0.3.1 can now also summarize your claude sessions into a journal like Obsidian.

[Soleone @soleone]

Are you losing track of how your time is spent every day across all these agent sessions?

I built logdig which writes daily summaries of your pi agent sessions into a directory, or hook it up with Obsidian.

Run one command to backfill all your history!

https://t.co/ltbaf1ddaG

[Soleone @soleone]

@theghost1623 Yea, the former, session transcript history, and not live tool calls.

[Soleone @soleone]

Are you losing track of how your time is spent every day across all these agent sessions?

I built logdig which writes daily summaries of your pi agent sessions into a directory, or hook it up with Obsidian.

Run one command to backfill all your history!

https://t.co/ltbaf1ddaG

[Soleone @soleone]

@lifeos_md Yes, when you click on a short summary timestamp it links to the details page which has the following info including the session id:

[Soleone @soleone]

@yawaramin @mitsuhiko It is correct. It is combining an assignment and if statement. But often it looks confusing so people opt to break it up into two lines. Don't even need the parentheses but regularly used to clarify an assignment instead of == check.

[Soleone @soleone]

@ardfard101 @mitsuhiko You can have pretty concise typing info in Ruby now as well. At the enterprise level we generally want typing for sanity.