Top all-time · #11203
Peter Sergeant
Personal blog · sgnt.aiaiprogramming
Peter Sergeant's personal blog at sgnt.ai covers ai and programming. Since 2025, 6 different Hacker News readers have linked it 37 times, most often “You could have built Jev”. It is still writing: the last post was on 2026-09-18, and you can follow it by feed.
Too few sampled comments (1) to say how often HN recommends it.?
Why HN links it
Only one comment cites the linked post 'Get the hell out of LLMs', asking whether that is the author's point about LLMs. The comment is a question and does not judge the writing.
Summarised by Claude from a sample of 1 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
a blog about LLMs, Agents, and AI, by Peter Sergeant
Most-linked posts?
- You could have built Jev3 people · 8 mentions · first linked 2026-09-18
- Understanding modern AI is understanding embeddings: a guide with almost no math2 people · 8 mentions · first linked 2025-05-08
- Don’t let an LLM make decisions or execute business logic2 people · 3 mentions · first linked 2025-04-01
What Hacker News says?
There are many, many, open-source versions of Jev, including three distinct projects sharing the name “openjev” If you’re interested in the basic trick most are using (which is probably also what Jev does) then it’s here: https://sgnt.ai/p/jev/
Comment linking /p/jev/
on “I built non-autoregressive decision models with RL a year ago”Submitted Don’t let an LLM make decisions or execute business logic as a story · 325 points
HN discussionNo idea at all what this particular project called openjev is doing, but https://github.com/TheoLeeCJ/openjev and https://github.com/ekzhang/openjev-sglang (neither of which I have any relation to) generate a single token, rather than JSON structured output. I wrote up this technique here: https://sgnt.ai/p/jev/
Comment linking /p/jev/
on “OpenJev”Submitted I will fix your vibe-coded MVP – sgnt.ai as a story · 19 points
HN discussionThat's the approach that daseinlabs/open-jev takes, in contrast to the above, which is what TheoLeeCJ/openjev and ekzhang/openjev-sglang do https://sgnt.ai/p/jev/
Comment linking /p/jev/
on “Jev in 25 Lines of Python”Submitted You could have built Jev as a story · 6 points
HN discussionYou need virtually no maths to deeply and intuitively understand embeddings: https://sgnt.ai/p/embeddings-explainer/
Comment linking /p/embeddings-explainer/
on “The maths you need to start understanding LLMs”Submitted LLM spambots liked my Show HN post more than real people did as a story · 5 points
HN discussionEmbeddings are reasonably simple, but it’s a journey to get there, and I am very proud of the dog-heavy explainer I wrote on them: https://sgnt.ai/p/embeddings-explainer/
Comment linking /p/embeddings-explainer/
on “RAG Is Simpler Than You Think”Submitted RAG chunking isn't one problem, it's three as a story · 4 points
HN discussionI wrote an embeddings explainer a few days ago if anyone is interested: https://sgnt.ai/p/embeddings-explainer/ Very little maths and lots of dogs involved.
Comment linking /p/embeddings-explainer/
on “Embeddings are underrated (2024)”Submitted When Users Won't Wait: Engineering Killable LLM Responses as a story · 4 points
HN discussionI wrote a simpler explanation still, that follows a similar flow, but approaches it from more of a "problems to solve" perspective: https://sgnt.ai/p/embeddings-explainer/
Comment linking /p/embeddings-explainer/
on “LLM Embeddings Explained: A Visual and Intuitive Guide”Submitted You could have built Jev as a story · 3 points
HN discussionEven that sounds too involved, just tell the agent that one exists, even if it doesn’t… https://sgnt.ai/p/terrible-mistake/
Comment linking /p/terrible-mistake/
on “Just the rumour of a bug is enough to find an exploit these days”Submitted Understanding modern AI is understanding embeddings: a guide with almost no math as a story · 3 points
HN discussionIt depends what you mean by agent, first of all, but I’m going to assume you mean what I’ve called “narrow agency” here[0]: “[an LLM] that can _plan and execute_ tasks that happen outside the chat window“. That humans make mistakes all the time is the reason we encode business logic in code and automate systems. An “if” statement is always going to be faster, more reliable, and have better observability than a human…
Comment linking /p/agentic-ai-bad-definitions/
on “Don’t let an LLM make decisions or execute business logic”Submitted In-memory free-text search is a super-power for LLMs as a story · 3 points
HN discussionJev has taught me the same lesson three times over now. When it first came out, I thought "this weekend, I'll do a little open-source Jev based on single-token prediction and the token logit output", but of course when it came to it, there were at least 5 that had already been done between me thinking that and getting around to it. So I wrote up[0] what other people had done, but wasn't happy with how weak the bench…
Comment linking /p/jev/
on “I turned Jev into a (lousy) chatbot”Submitted You could have built Jev as a story · 2 points
HN discussionYeah, that's absolutely a real phenomenon, but I don't think it's especially OpenClaw-specific: I have lots of folders that are just collections of instructions and scripts that have mostly started with a prompt, and that I run sandboxed agents over. Previously I'd have considered automating them, but now it's just "fire up sandbox in a project folder, let the tool build itself". I wrote about this a bit here: https…
Comment linking /p/the-software-i-stopped-writing/
on “OpenClaw 2.0, Accidentally”Submitted Maybe software stops being the default as a story · 2 points
HN discussionIt is obvious. I was going to build my own little toy doing just that ten days ago, and then found four pre-existing projects, so wrote those up instead: https://sgnt.ai/p/jev/
Comment linking /p/jev/
on “Turning GLM-5.3-Flash into a Jev-like decision model”Submitted Chat, we've made a terrible mistake as a story · 2 points
HN discussionYes! I just wrote an article on this: https://sgnt.ai/p/hell-out-of-llms/
Comment linking /p/hell-out-of-llms/
on “LLM Workflows then Agents: Getting Started with Apache Airflow”Submitted Four bad definitions of "Agentic AI" as a story · 2 points
HN discussionHere is one I wrote today on LLMs that can handle chat input like humans write: multiple disjointed messages arriving asynchronously that need to be treated as one: https://sgnt.ai/p/interruptible-llm-responses/ I use a similar technique to the article for trying to avoid jailbreaks by putting untrusted input through zod to check I got back a JSON structure of the right shape, which has been very effective. I’ve bee…
Comment linking /p/interruptible-llm-responses/
on “CaMeL: Defeating Prompt Injections by Design”Submitted The prompt is not the work; describing AI contributions as a story · 1 points
HN discussionThe fallacy here is that you can reduce the wide range of things that “AI generated” means into a single flag. As for why not “show us the prompt”: https://sgnt.ai/p/prompt-is-not-the-work/
Comment linking /p/prompt-is-not-the-work/
on “Ask HN: Add flag for AI-generated articles”Submitted Understanding Modern AI Is Understanding Embeddings: A Guide for Non-Programmers as a story · 1 points
HN discussion
