HNHacker News
TopNewBestAskShowJobs

pietz

391 karma · joined December 2, 2016

submissionscomments
pietz··on Disrupting the first reported AI-orchestrated cyber espionage campaign
I imagine Anthropic employs a lot of talent from China. Beyond the political, they should be fairly certain to publish these claims to avoid an internal shit storm.
pietz··on Why agents do not write most of our code – A reality check
Can you go into a bit more detail why the two approaches are so different in your opinion?

I don't think I agree and I want to understand this argument better.

pietz··on GPT-5-Codex-Mini – A more compact and cost-efficient version of GPT-5-Codex
Since they also limited the usage of codex CLI quite a bit, this might help. I really like the gpt-5-codex model. It has the most impressive dynamic reasoning effort I've seen. Responding instantly to simple questions while thinking very long when necessary. It's also interesting how close it is in many benchmarks to gpt-5. It's not just a good coding model. It's a great agentic model.
pietz··on Studio Ghibli, Bandai Namco, Square Enix Demand OpenAI to Stop Using Their IP
Thanks for sharing the font story! That made me quite a bit happier than it should have.
pietz··on Production RAG: what I learned from processing 5M+ documents
Cannot confirm this. Both sound like prompting issues.

- Depends on your use case to let the model understand when and when not to use tools - gpt-5 s VERY persistent and often searches more than 10 times in a single run depending on the results.

We're using pydantic AI where the entire Agent loop is taken care of by the framework. Highly recommend.

pietz··on Show HN: Deta Surf – An open source and local-first AI notebook
Oh, it's nothing fancy I'm afraid.

I have an AGENTS.md file in my vault that provides a bit of context that this is not about coding, but about serving as my Obsidian helper. I ask it to check the .obsidian folder to check my plugins, Tell it how I like to build my presentations and the workflow I follow when writing articles.

Then I just launch codex in the terminal inside the vault, have it do stuff, while I watch the desktop app for change.

Other coding agents like Claude Code should work also, but gpt-5-codex was specifically trained on using the terminal to do everything it and doesn't even have that big of a system prompt related to coding. Works well.

Combine that with a speech-to-text app and codex blazes through to find stuff or do things for me.

pietz··on Show HN: Deta Surf – An open source and local-first AI notebook
It has been wild watching deta over the years.

They didn't pivot, they completely reinvented themselves. Twice.

I loved their first cloud offering, which they sadly abandoned.

Then they launched Space, which was kinda cool, but mostly weird and raised the question "why?". Also cancelled.

Surf looks mostly cool, although I also don't quite understand it. It seems like Notion with a different twist on AI. Not sure. Since I'm fairly happy with my Obsidian + Codex setup, I'll pass for now. The good news is, this one's open source!

I'd love to know how they're financing all of this. They have been around for years and users never even had the option to drop money in their lap. Now they're trying open source. Wild ride.

All the best!

PS: I would have paid for deta cloud Pro ;)

pietz··on My trick for getting consistent classification from LLMs
Makes sense, I appreciate the comment. Well written article. Subscribed.
pietz··on My trick for getting consistent classification from LLMs
I enjoyed reading this, but it seems overly complex and at least slightly flawed.

Why not embed all tweets, cluster them with an algorithm of your choice and have an LLM provide names for each cluster?

Cheaper, better clusters and more accurate labels.

pietz··on Production RAG: what I learned from processing 5M+ documents
My biggest RAG learning is to use agentic RAG. (Sorry for buzzword dropping)

- Classic RAG: `User -> Search -> LLM -> User`

- Agentic RAG: `User <-> LLM <-> Search`

Essentially instead of having a fixed loop, you provide the search as a tool to the LLM, which does three things:

- The LLM can search multiple times

- The LLM can adjust the search query

- The LLM can use multiple tools

The combination of these three things has solved a majority of classic RAG problems. It improves user queries, it can map abbreviations, it can correct bad results on its own, you can also let it list directories and load files directly.

pietz··on Production RAG: what I learned from processing 5M+ documents
But the model is like 18 months old. and recently we've seen big leaps on MTEB. Not sure how well those translate to reality, but I'm a little surpised this wasn't worth looking into.
pietz··on Production RAG: what I learned from processing 5M+ documents
I find it interesting that so many services and tools were investigated except for embedding models. I would have thought that's one of the biggest levers.
pietz··on DeepSeek OCR
I threw in the first image/table into Gemini 2.5 Pro letting it choose the output format and it looks like it extracted the data just fine. It decided to represent the checkboxes as "checked" and "unchecked" because I didn't specify preferences.
pietz··on DeepSeek OCR
Maybe my imagination is limited or our documents aren't complex enough, but are we talking about realistic written documents? I'm sure you can take a screenshot of a very complex spreadsheet and it fails, but in that case you already have the data in structured form anyway, no?
pietz··on DeepSeek OCR
Could you provide an example that fails? I'm interested in this.
pietz··on DeepSeek OCR
We ran a small experiment internally on this and it looked like Gemini is better at handwriting recognition than I am. After seeing what it parsed, I was like "oh yeah, that's right". I do agree that instead of saying "Sorry, I can't read that" it just made up something.
pietz··on DeepSeek OCR
My impression is that OCR is basically solved at this point.

The OmniAI benchmark that's also referenced here wasn't updated with new models since February 2025. I assume that's because general purpose LLMs have gotten better at OCR than their own OCR product.

I've been able to solve a broad range of OCR tasks by simply sending each page as an image to Gemini 2.5 Flash Lite and asking it nicely to extract the content in Markdown under some additional formatting instructions. That will cost you around $0.20 for 1000 pages in batch mode and the results have been great.

I'd be interested to hear where OCR still struggles today.

pietz··on Claude Haiku 4.5
I think with gpt-5-mini and now Haiku 4.5, I’d phrase the question the other way around: what do you need the big models for anymore?

We use the smaller models for everything that’s not internal high complexity tasks like coding. Although they would do a good enough of a job there as well, we happily pay the uncharge to get something a little better here.

Anything user facing or when building workflow functionalities like extracting, converting, translating, merging, evaluating, all of these are mini and nano cases at our company.

pietz··on Figure 03, our 3rd generation humanoid robot
Give me a 30min uncut video of it doing chores around the house to make me a believer.
pietz··on Vibe engineering
Simon went quickly from not agreeing with common buzz words to making up his own.
pietz··on Which table format do LLMs understand best?
This article screams for a accuracy vs. tokens plot. Thanks though, interesting results.
pietz··on GPT-5 Codex: How it solves for GPT-5's drawbacks
The thing that interests me the most about gpt-5-codex is the fact that it thinks less in easier situations and more in complex situation, without changing the reasoning effort. If OpenAI can double down on this, they may have found a better solution to their model routing dilemma.
pietz··on The AI coding trap
Interesting take.

I think it boils down to personal preference where some people want to use AI while others don't. I also learn when coding with my AI agent. I learn about using the tool more effectively. As someone who has been coding for 10 years, I find more pleasure in AI assisted coding.

But aside from taste, the product and the business don't care about what I like. It's about shipping quality updates more quickly. And while there might be some tension in saying this, I'm convinced that I can do that much more quickly in AI assisted coding.

pietz··on The AI coding trap
I don't know how the author uses AI coding tools, but all agents I've been using recently (Codex and Claude Code) do planning and exploration before writing code. I don't even know how it would work the other way around.

I do see different behavior whether planning or exploration comes first though.

pietz··on GPT‑5-Codex and upgrades to Codex
Thank you for your work, Simon.
pietz··on GPT‑5-Codex and upgrades to Codex
Do we really know that gpt-5-codex is a finetune of gpt-5(-thinking)? The article doesn't clearly say that, right?

I suspect that this is smaller than gpt-5 or at least a quantized version. Similar to what I suspect Opus 4.1 is. That would also explain why it's faster.

pietz··on Show HN: Base, an SQLite database editor for macOS
How have I never heard of this before? Been looking for a sleek, minimal SQLite client for more than a year now.
pietz··on AGENTS.md – Open format for guiding coding agents
Holy sh*t, I accidentally for got the "S" and was directly forwarded to PayPal.
pietz··on AGENTS.md – Open format for guiding coding agents
This isn't a format. It's just a convention that literally boils down to: Put LLM instructions in a text file and call it AGENTS.md.

Well, thanks I guess?

pietz··on Show HN: Optimization for LLM App
This sounds a bit backwards to me. Instead of everyone improving their websites, I expect LLMs will get better at fetching and extracting. (It really doesn't take much).
← PreviousPage 4 of 6Next →