@jamesm If you ask your mom "is the sky blue" she says "yes" immediately, that's jev. If you ask your dad "why is the sky blue" he talks for ten minutes, that's an LLM.↗

@rcx86 If only I had thought to train it on StackOverflow instead I think it would have clicked for me that LLMs can be promptable, general purpose Q&A engines. Ah well.↗

Built a distributed DNS server in ~a week. Rebuilt twice on purpose. Shipped -> went on vacation. DNS incidents a month later: 0. How we did it (hint: the answer isn't vibe-coding): blog.exe.dev/claude-is-not-…↗

recommended reading. don't get turned off by the title. this is how i use LLMs too for many tasks. jamesdrandall.com/posts/thrust_a…↗

Fable turned my remarkable into Tom Riddle's diary from Harry Potter. My prompts fade, a LLM respond. Magical!↗

Introducing PorTAL: Portable Task Adapters for LLMs. A novel recipe to cheaply port fine-tuning between models. It matches per task LoRA accuracy at half the cost, lowering the switching overhead of adapting tasks across LLMs. At Ramp, every new model release used to mean retrai…↗

Gemini 3.5 Flash is built to help you execute complex, agentic workflows. 3.5 Flash rivals flagship models to deliver frontier performance for agents and coding, at the lightning speeds you expect from the Flash series.↗

wait, prompts are code, files are state? YC catching up to ca. Q2 2025 mariozechner.at/posts/2025-06-…↗

Introducing SubQ - a major breakthrough in LLM intelligence. It is the first model built on a fully sub-quadratic sparse-attention architecture (SSA), And the first frontier model with a 12 million token context window which is: - 52x faster than FlashAttention at 1MM tokens -…↗

Today, we're shipping the new Resend email editor, extended by AI. What I love about this is that you can BYOA (Bring Your Own Agent). There's a built-in LLM that can help you draft an email based on an URL. But you can just MCP into the editor too.↗

When Opus 4.5 came out, it was a one-way door to a new way of engineering. Agents now do most of our coding. Knowing the inherent flaws and over-confidence of LLMs, we sent a clear message to our teams. Vibing and mission-critical infrastructure don’t go together. We’re sharing…↗

- Drafted a blog post - Used an LLM to meticulously improve the argument over 4 hours. - Wow, feeling great, it’s so convincing! - Fun idea let’s ask it to argue the opposite. - LLM demolishes the entire argument and convinces me that the opposite is in fact true. - lol The LLM…↗

anytime i finish a blog post, i feed it to an LLM asking it to produce 20-40 HN or Reddit comments. immensely effective. stole that idea from @mitsuhiko↗

openwork just reached 100k downloads. but what is openwork, exactly? openwork is an app that lets you share ai agents with friends and coworkers. we call them "workers." at minimum, a worker is just an llm with access to your filesystem and tools - think claude code/codex-sty…↗

I'm thinking about this a lot lately — did we only ever create abstractions to avoid duplication, and now is maintaining duplication actually not really a problem? I'm working on this very componentized marketing site kit product we're building and can't shake the feeling that ma…↗

Don't think of LLMs as entities but as simulators. For example, when exploring a topic, don't ask: "What do you think about xyz"? There is no "you". Next time try: "What would be a good group of people to explore xyz? What would they say?" The LLM can channel/simulate many pe…↗

Kudos to @AnthropicAI for a killer registry of Claude use cases. If you're struggling for inspiration for what to do with an LLM (any LLM) this is a great set of scenarios to sift through. claude.com/resources/use-…↗

Talked to @willccbb turns out it’s really just what I thought it was. Agent with a submit for review function. Agent is jsut an llm in a form loop. Review function is just another agent with a rubric that approximates a loss function.↗

1/ Base44 ended up making $189K *profit* in May - not $100K as expected. What changed: – Faster, unexpected growth – Drastically reduced LLM costs (see cost graph below) Sharing some details on what helped improve margins 🧵↗

Introducing apple-mcp - git.new/apple-mcp one simple command to give LLMs access to a bunch of apple-native tools like - contacts - notes - iMessages and more (soon) just add this to your claude desktop config. done.↗

i’ve been messing around with @duolingo a lot lately—not because i had some grand ambition to become fluent in another language, but because i thought it might impress someone (spoiler: she’s gone now). in my mind the most interesting thing though is what duolingo does for lang…↗

Have been playing with OpenAI's "Operator" agent all weekend. Some early thoughts: 1) The fact that it's already hosted and you can just start using it means it'll be more widely used than Claude's "Computer Use" feature. 2) Operator will be able to make websites that haven't…↗

Recently, I was at the library & stood in front of a book that might be a good read. I fired up Gemini, asked it for a plot summary & a digest of the reviews before deciding to read it. So I started a list : what are the ways I use AI that are new in the last year? I often publ…↗

LLM Fine tuning is here! San Francisco’s top AI engineers came together to see what’s possible with fine-turning and only 4 hours of hacking. Here’s an exclusive what we saw at the “Anything But Wrappers” hackathon (🧵):↗

To: My fellow data/info-biz nerds This is a great summary and another reason why proprietary quantitative or qualitative data & insights are going to be so powerful in this generative AI / LLM future h/t @joeobrien513↗

🗄️🗄️Will the next generation of LLMs have native data structures for storing insights? Pre-LLM era research has some interesting ideas here. Two of my favorite, both exactly what they sound like: - Stack-Augmented Recurrent NNs - Neural Turing Machine More 👇↗