Krisjanis Gaidis (@kgaidis) header
Krisjanis Gaidis (@kgaidis) profile picture

hi! ❤️ boost your focus w/ Refocus - link in bio (100k MAU) ex: @stripe @meta @adobe

Santa Monica, CA
Joined January 2010
Photo by Krisjanis Gaidis (@kgaidis)Photo by Krisjanis Gaidis (@kgaidis)Photo by Krisjanis Gaidis (@kgaidis)Photo by Krisjanis Gaidis (@kgaidis)Photo by Krisjanis Gaidis (@kgaidis)Photo by Krisjanis Gaidis (@kgaidis)Photo by Krisjanis Gaidis (@kgaidis)Photo by Krisjanis Gaidis (@kgaidis)Photo by Krisjanis Gaidis (@kgaidis)Photo by Krisjanis Gaidis (@kgaidis)
I block non-productive websites 24/7 by default, so when my mind autopilots to a random website, I have to pause an extra second to ask if I truly intended the context switch.
1
1
1,113
"Most aim for 100% accuracy, and they end up with 50% clarity. Aim for 100% clarity, 80% accuracy." Very interesting. Without this, the model is not displaying "intelligence." It's displaying "LLM output." For example, I just got this notification from my Dot: "Refocus 2.2.11 now appears as the latest version on the App Store." This was Dot predicting what's important to me. It's not being "intelligent." It's blindly outputting information that is obvious to me: I pressed the button to release, and it's been released for several hours where I even have a thread monitoring the release. If the model can't discern what's important, how can it make decisions? (ex. proper code architecture for Y problem)
I took this to heart and spent the last few hours doing another deep dive on GPT-6.2 Astra tl;dr - things are better than when I tested, but IMO still "directionally, very bad" I'm going to keep using the same example I've hyperfixated on since the start: "why is creating a worktree so slow?" I had a long thread trying to diagnose perf issues while using t3 code on a particular VPS. 2 of the 3 were fast, 1 was really slow. I asked astra outright "why is this so slow?", and it said a bunch of gibberish. I followed up with "I feel like I'm going insane, can you just tell me the cause of the slowness?" I'll give you the answer upfront: this VPS uses a virtual disk that is slow as shit, my other VPSs have physical NVME drives that are faster. Image 1 is from before the fixes Tibo mentioned. It shows an unnecessary graph, focuses too much on pnpm and file systems, bolds irrelevant text, and brings up ram just as much as it (subtly) mentions drive speed. Image 2 is of Opus asked same thing. It opened with the important part in bold: "Slow disk, made worse by VDO". Ideal! Image 3 is 6.2 Astra post-fixes Tibo mentioned. Better? Absolutely. Much more concise. Sadly, it still focuses way too much on second-order problems that are a small % of the cause. "XFS over a compression and deduplication layer on a virtual disk." - so close, but still a miss. I think this perfectly encapsulates the problem I have with 6.2 Astra. It doesn't miss anything. It's SO thorough. But in it's thoroughness, I find it misses the most important thing: the actual point. It wants to know everything, and then it wants to share everything. It doesn't understand what is and isn't important. Did it say anything wrong in my screenshots? Not really. But IMO, it still entirely failed to miss the point of my question, and gave an answer that was too focused on accuracy and not enough on clarity. In the wise words of Michael Seibel: "Most aim for 100% accuracy, and they end up with 50% clarity. Aim for 100% clarity, 80% accuracy". Silly way of putting it, but I genuinely wish OpenAI models would lie to me a LITTLE more so I could better understand the vibe, instead of having my time wasted by all these useless details.
1
1
43
with Claude/Codex, I can now design anything I wish without any technical ability ...but that is not enough. the "alpha" is in the details. you could say the same thing about vibe coded apps: even if anyone can make anything, it doesn't mean downloads, retention and profits
the bitter lesson of App Store Screenshots is that you really gotta get the details right and a "pretty design" won't just mean a clear conversion boost
1
1
78
the bitter lesson of App Store Screenshots is that you really gotta get the details right and a "pretty design" won't just mean a clear conversion boost
1
1
118
based on reviews of new OpenAI models, got a Claude sub for double checking work+architecture - too important to not use the smartest models wild that this release boosts Anthropic IPO
29
finally. I'm so happy. bonus: "Pro $200 subscribers can use GPT-6.2 Sol in Ultrafast mode"
Codex update: GPT-6.2 Astra and GPT-6.2 Sol in Work and Codex What changed: • GPT-6.2 Astra and GPT-6.2 Sol are now available in ChatGPT Work and Codex. • Astra is for harder projects; Sol is for everyday coding, writing, and multistep workflows. • GPT-6.2 Sol lets you complete more tasks on the same paid plan than GPT-6.1 Sol. Details in thread.
1
79
"the worst model OpenAI has ever released" 🥲 teapawt.net/theo/status/2122757647…
6.2 Astra is here. I’ve been testing it for awhile. I think this is the worst model OpenAI has ever released. I’ll share more soon.
1
21
"they were essentially unusable" teapawt.net/mattshumer_/status/212…
I will not be writing a full review of GPT-6.2 Astra / Sol. I had a terrible experience testing them. They were essentially unusable. I do not recommend these models.
6
interesting that everyone is converging on the bots paradigm
Meet Devin Bots. From writing firmware to merging thousands of PRs a week, a Devin Bot leads a team of agents to get the job done.
17
reminds me of what @markgurman once said: "if you think about how much money the App Store makes, and what percentage of the App Store is casino type of games, it’s billions"
I reverse engineered monopoly go
1
52
Apple makes more money from casino games than it does from selling Macs, according to @markgurman. "If you think about how much money the App Store makes, and what percentage of the App Store are these casino games, it's billions."
8
Episode #303: Rose Blumkin (Warren Buffett's Favorite Founder) is likely the most underrated @FoundersPodcast episode
25
if Elon Musk has to defend his accomplishments then it’s a futile exercise for anyone
Let’s take this gentleman’s argument at face value that anyone can throw money at rockets and get something like Starship, Falcon or Dragon. Perhaps he can explain to everyone how SpaceX is able to catch giant starships from orbit with chopsticks and deliver 85% of all Earth payload to orbit with Falcon, while Blue Origin, which started before SpaceX and was funded by the immensely wealthy Jeff Bezos, is not yet able to launch reliably? Also, why is it that Boeing and Lockheed received several times what SpaceX did from the government for development of their rockets and spacecraft, but the Boeing capsule is still not operational and ULA delivers less than 5% of SpaceX payload to orbit? I am genuinely curious about these questions.
1
2
200
it’s like Michael Jordan convincing others that he’s, actually, good at basketball
1
27
There's currently an Sign in with Apple incident Domain: com.apple.AuthenticationServices.AuthorizationError Code: 1000 status.claude.com/incidents/…
1
1
361
Refocus still has a lot more iPhone Air users than iPhone Duo (13x)
It crossed over today. Overcast now has more users on iPhone Duo than iPhone Air. It took nearly three weeks, but that's still pretty impressive.
1
115
~51% of Duo users are paying subscribers, ~21% of Air are
18
interesting pattern: I use Figma now more than ever because it's so easy to use with agents
30
even my X Ads are about Jensen
29