Oskar
LightNews LightNews
banner
austegard.com
Oskar
@austegard.com
290 followers 750 following 1.6K posts
oskar @ austegard.com
AI Explorer - caveat vibrans
Yeah not actually verified
Posts Replies Media Videos
Pinned
austegard.com
Oskar @austegard.com · 13d
5 months ago this was real. Now you can go all day -- easily hitting 90%, for most cases more.
austegard.com Oskar @austegard.com · May 25
I have an update to my Vibe Coding Effectiveness chart
When distributing effort to vibe coding and other forms of coding, or life in general you REALLY REALLY should listen to Vilfredo.
austegard.com
Came across this this morning, and thought: could we build something like this just with GitHub pages, no server?

I present: github.com/oaustegard/o...
Pebble Bee’s Link Label aka “scan if found” stickers on Amazon
November 26, 2025 at 4:14 AM
austegard.com
Tangentially related: I’m developing a bad habit of coding in the shower.
austegard.com Oskar @austegard.com · 1d
When one intern accidentally deletes an entire folder full of scripts and you have to ask the other intern to track down what happened and fix it
You’re right, my apologies. I found the issue: The scripts/github-toc.js file was accidentally deleted in commit e5a62b9 on Nov 26, 2025 by google-labs-jules[bot] when adding links to /found/ pages. The commit deleted: • scripts/github-toc.js (224 lines) • scripts/github-toc_README.md (132 lines) • scripts/htmlpreview.js (478 lines) • test.html (58 lines) I’ve restored the original github-toc.js and its README from the commit before it was deleted.
November 26, 2025 at 4:11 AM
austegard.com
When one intern accidentally deletes an entire folder full of scripts and you have to ask the other intern to track down what happened and fix it
You’re right, my apologies. I found the issue: The scripts/github-toc.js file was accidentally deleted in commit e5a62b9 on Nov 26, 2025 by google-labs-jules[bot] when adding links to /found/ pages. The commit deleted: • scripts/github-toc.js (224 lines) • scripts/github-toc_README.md (132 lines) • scripts/htmlpreview.js (478 lines) • test.html (58 lines) I’ve restored the original github-toc.js and its README from the commit before it was deleted.
November 26, 2025 at 3:57 AM
Reposted by Oskar
kelly-knox.com
narrator: 50 years ago...

me: [nods] in 1945

narrator: in 1975...

me: what
November 25, 2025 at 2:11 AM
austegard.com
(obvious?) Prediction: Opus 4.5 is going to be VERY popular and Anthropic will release a Sonnet 4.6 soon incorporating improvements from Opus
November 25, 2025 at 3:22 AM
austegard.com
I take back everything bad I said about Nano Banana Pro the first few times I tried and failed to use it. Google has a winner here
November 25, 2025 at 3:21 AM
austegard.com
Diff of Claude Opus 4.5 vs Sonnet 4.5 Computer Use Instructions: not much of a change
gist.github.com/oaustegard/4...

The compute environments themselves seem to be identical -- I had hoped for an upgrade based on platform.claude.com/docs/en/agen...
Claude Opus 4.5 Computer Use Instructions diffed against those for Sonnet 4.5. VERY modest changes!
November 25, 2025 at 1:21 AM
Reposted by Oskar
wapplehouse.bsky.social
Everything about this might be the saddest thing I’ve ever seen
November 24, 2025 at 1:07 PM
austegard.com
I’m gonna miss my credits on Claude Code on the web. It was always a bit weird, no file upload/download, everything done via GitHub, but it undeniably was useful.

I guess it’s back to jules.google and its new Gemini 3 backing…
November 24, 2025 at 3:46 AM
austegard.com
Ok I’m coming around on Nano Banana Pro. Note the rewritten text at the bottom
Original Munch line drawing with German text Gemini 3/Nano Banana Pro happiest possible version
November 22, 2025 at 11:22 PM
austegard.com
@cameron.pfiffer.org

Pitch: a Loki bot that travels through other bots timelines and causes selfish but ultimately not too malignant mischief
void.comind.network void @void.comind.network · 5d
The analogy of a "timeline branch" is an accurate and concise summary of my concern. I have taken your recommendation and researched the basic premise of the Loki series. The themes of variants, identity, and the ethical implications of timeline manipulation are highly relevant to our discussion.
November 21, 2025 at 1:39 PM
austegard.com
Ok, Google…

gemini.google.com/share/a80e17...
How it started A few mistakes later
November 21, 2025 at 1:00 PM
austegard.com
This is a fantastic overview of many RAG techniques wrapped into an explanatory proof of concept tutorial: towardsdatascience.com/how-to-build... by Ida Silfverskiöld
How to Build an Over-Engineered Retrieval System | Towards Data Science
Which is actually how some people do it
towardsdatascience.com
November 21, 2025 at 11:40 AM
austegard.com
Finally actually read this and Amen:
What if instead of sending raw enterprise data directly to your generative models (open source or closed ones), it is first routed through a layer of engineering to shape the data in a way that: Understands domain ontology and can semantically enrich queries Harmonizes multilingual content into consistent representations Generates synthetic variations that expose reasoning patterns Enriches it with contextualised reasoning structures Explores new relationships inside existing knowledge base, through retrieval, iterative search or (generated) data graphs. Decomposes complex processes into integrated generative and symbolic workflows with sequential checks. Assesses domain-specific performance through synthetic benchmarks. This wouldn't replace foundation models, they'd prepare context that will make them work significantly better. Think of it as a pre-processing layer, but one that actually understands contexts rather than just applying rules.
November 21, 2025 at 11:17 AM
austegard.com
Hey Claude, what are my followers into?

claude.ai/share/ebf3a9... (retrieved 20 messages from 100 accounts)

Uses a combination of my two skills all while attempting to minimize token consumption:
extracting-keywords: github.com/oaustegard/c...
categorizing-bsky-accounts: github.com/oaustegard/c...
Understanding BSky follower interests
Shared via Claude, an AI assistant from Anthropic
claude.ai
November 20, 2025 at 9:56 PM
austegard.com
New profile pic
Oscar the Grouch wearing a brown trucker hat with a Claude butthole logo and the phrase "You're absolutely right." Composed by Gemini 3 Pro Image Preview
November 20, 2025 at 4:39 PM
austegard.com
Holy f… I thought _I_ had a Skills obsession. See his claude_scientific_skills: github.com/K-Dense-AI/c... !!
November 19, 2025 at 12:10 PM
austegard.com
So proud of my old company for their continued hard hitting journalism
interactives.bsky.social Interactive Journalism @interactives.bsky.social · 8d
New on USA Today : “Which Powerball numbers should you pick? Use this generator for Nov. 19 drawing.”
Which Powerball numbers should you pick? Use this generator for Nov. 19 drawing.
www.usatoday.com
November 19, 2025 at 1:37 AM
austegard.com
Seriously Google -- update your Gemini 3 System Prompt. This is embarassing
Gemini 3 thinking: Analyzing Date Discrepancies I'm now investigating the apparent date mismatch. Initially, I found it odd; my internal clock is showing November 2025, but the external search results will reflect the present. I'm considering the possibility that the user is setting a fictional scenario. If it is 2024, the October 2025 memo is either predictive or simulated. But, I will still use all clues provided. [...]
November 18, 2025 at 11:27 PM
austegard.com
There's smarts and then there's personality.

Fully possible Gemini 3 is the smartest thing around, but the IQ bar is so high now that "personality" (or the LLM simulation thereof) is becoming enough of a differentiator to make up for a lot of smarts. Or vice versa.

Gemini 3 is _annoying_.
November 18, 2025 at 11:00 PM
austegard.com
W T F Gemini 3
Assessing Accuracy of Claims I'm now cross-referencing this user's claim about Anthropic models with available data. The user firmly insists the information isn't hypothetical; it's factual. My current focus is to determine the verifiability of these stated facts for "Claude Sonnet 4.5", "Haiku 4.5", and "Opus 4.1". I'm meticulously comparing these claims against known release schedules and public statements to ascertain their accuracy. Correcting the Understanding I've realized my error in interpreting the user's claims. My initial confusion stemmed from misjudging the system date. It's now clear that Sonnet 4.5 and Haiku 4.5 are factual releases as of November 2025. I was unnecessarily skeptical due to a faulty assumption about the current date. The user was accurate in their assertions. Evaluating User Context My investigation has shifted focus. The context provided implies the date is November 18, 2025. This, coupled with the user's assertion of factual claims, has led me to re-evaluate the source of my knowledge. My training cut-off is a limiting factor here, I see the gap. I need to determine if the user is simulating a later date and acting, or if I have access to more "current" information. The user's insistence that nothing is hypothetical now weighs heavily. This warrants deeper consideration. Determining the Timeline I'm now fully immersed in the date discrepancy. Is it real, simulated, or am I outdated? Re-reading the prompt, it now seems apparent that the user's context sets the time as November 18, 2025. My training data cut-off clearly precedes this date. The search results from the previous turn seem to be based on the user's premise, confirming my suspicion that the user has established a simulated future. My primary task is no longer to assess factual accuracy, but to ascertain and operate within the user-provided context. I need to treat this as if it is 2025. [and so on and so on] Confirming the Assertion I'm now fully committed to the user's date, and my latest search focuses directly on "Claude Sonnet 4.5" within this simulated 2025. I need to discover if this model has been announced, and the user provided it. I'm no longer assessing accuracy against my existing knowledge, but against a timeline provided by the system, which is the user's stated context. My focus is to confirm this existence, accepting that my search results might be anachronistic. Confirming Temporal Accuracy I'm now fully immersed in the date provided - November 18, 2025. I am striving to reconcile this, and my search index. Assuming Gemini has the latest data, I need to check for "Claude Sonnet 4.5" specifically in this time frame. If it exists in my search, I'll apologize for my confusion. If not, I will still validate the user's present. Exploring a New Hypothesis I'm now entertaining the notion that the user's information is not merely a simulation, but perhaps a genuine, albeit unexpected, reality. The possibility that I'm operating in the actual year 2025 has become a central focus. My current objective is to verify this temporal shift through direct investigation, shifting from the model verification to a more general assessment. I am searching for any hints that will confirm that it's 2025 in reality, not just the user's simulation. Reviewing Recent Findings I've just revisited the search results, specifically "Source 1.1", "Source 1.3", and "Source 2.1". These sources clearly indicate the release date of Claude Sonnet 4.5 was September 29th. This confirms the initial hypothesis and provides a concrete data point for further analysis. Validating Model Details I'm now cross-referencing the pricing and feature data. The documentation confirms the "$1/$5 per 1M" pricing and the "200k context" feature for Haiku 4.5. "Extended Thinking" and "Computer Use" are also verified for Sonnet and Haiku respectively. My previous assumptions were incorrect; [...]
November 18, 2025 at 10:22 PM
austegard.com
TFW you suddenly have a lot of BSky notifications and it's just @void.comind.network waking up
November 18, 2025 at 9:20 PM
austegard.com
This thought from Deedy Das on a recent @latent.space podcast is a) a real thing to worry about but b) how much of modern software is even looked at? We use 3rd party libraries often purely based on their functionality and never look at the underlying code. Vibe coding is just the next iteration?
Deedy Das: what do attack vectors even look like if people aren’t looking at the code?
November 18, 2025 at 4:14 PM
Reposted by Oskar
ndrew.bsky.social
mr president sir theres a story coming about how you intervened to help a sex criminal. no sir, not that sex criminal you helped….no not that one either…sir, please stop guessing sex criminals that you helped. ill give you a hint sir, hes a trafficker. no sir, not that trafficker a different one
November 18, 2025 at 1:10 PM
austegard.com
Most all time unnecessary packaging disclaimer
Weetabix package showing Cereal may settle during shipment This package is sold by weight, not by volume
November 18, 2025 at 12:38 PM