Fast

Our $100M Series B

Vibe code is legacy code

Crush: Glamourous AI coding agent for your favourite terminal

The HTML Hobbyist (2022)

Optician Sans – A free font based on historical eye charts and optotypes

Ollama's new app

Problem solving often a matter of cooking up an appropriate Markov chain (2007)

I think I&#x27;m officially a side project collector.I&#x27;ve had it all:A SaaS for freelancers... that I never had time to finish because I&#x27;m a freelancer.A revolutionary AI tool that I abandoned as soon as GPT-4 came out.And the famous &quot;anti-social media social network&quot; (spoiler: it was just me).I buy a domain name → I code for 3 all-nighters → I lose interest → I start again.My Google Domains look like a graveyard of unfinished dreams.But honestly, I&#x27;ve never learned so much, nor enjoyed it so much.And one day, I might release one that takes off. Or not. But I&#x27;ll be ready.Any other serial side-projectors here? Share your greatest fails&#x2F;unlikely successes

I launched 17 side projects. Result? I'm rich in expired domains

Australia widens teen social media ban to YouTube, scraps exemption

Writing memory efficient C structs

The math is haunted

A major AI training data set contains millions of examples of personal data

Figma will IPO on July 31

Critical vulnerability in AI coding platform Base44 allowing unauthorized access

Hi HN, we’re Abhinav, Andy, and Jeremy, and we’re building Lucidic AI (<a href="https:&#x2F;&#x2F;dashboard.lucidic.ai">https:&#x2F;&#x2F;dashboard.lucidic.ai</a>), an AI agent interpretability tool to help observe&#x2F;debug AI agents.Here is a demo: <a href="https:&#x2F;&#x2F;youtu.be&#x2F;Zvoh1QUMhXQ" rel="nofollow">https:&#x2F;&#x2F;youtu.be&#x2F;Zvoh1QUMhXQ</a>.Getting started is easy with just one line of code. You just call lai.init() in your agent code and log into the dashboard. You can see traces of each run, cumulative trends across sessions, built-in or custom evals, and grouped failure modes. Call lai.create_step() with any metadata you want, memory snapshots, tool outputs, stateful info, and we&#x27;ll index it for debugging.We did NLP research at Stanford AI Lab (SAIL), where we worked on creating an AI agent (w&#x2F; fine-tuned models and DSPy) to solve math olympiad problems (focusing on AIME&#x2F;USAMO); and we realized debugging these agents was hard. But the last straw was when we built an e-commerce agent that could buy items online. It kept failing at checkout, and every one-line change, tweaking a prompt, switching to Llama, adjusting tool logic, meant another 10-minute rerun just to see if we hit the same checkout page.At this point, we were all like, this sucks, so we improved agent interpretability with better debugging, monitoring, and evals.We started by listening to users who told us traditional LLM observability platforms don&#x27;t capture the complexity of agents. Agents have tools, memories, events, not just input&#x2F;output pairs. So we automatically transform OTel (and&#x2F;or regular) agent logs into interactive graph visualizations that cluster similar states based on memory and action patterns.
We heard that people wanted to test small changes even with the graphs, so we created “time traveling,” where you can modify any state (memory contents, tool outputs, context), then re-simulate 30–40 times to see outcome distributions. We embed the responses, cluster by similarity, and show which modifications lead to stable vs. divergent behaviors.Then we saw people running their agent 10 times on the same task, watching each run individually, and wasting hours looking at mostly repeated states. So we built trajectory clustering on similar state embeddings (like similar tools or memories) to surface behavioral patterns across mass simulations.We then use that to create a force-directed layout that automatically groups similar paths your agent took, which displays states as nodes, actions as edges, and failure probability as color intensity. The clusters make failure patterns obvious; you see trends across hundreds of runs, not individual traces.Finally, when people saw our observability features, they naturally wanted evaluation capabilities. So we developed a concept for people to make their own evals called &quot;rubrics,&quot; which lets you define specific criteria, assign weights to each criterion, and set score definitions, giving you a structured way to measure agent performance against your exact requirements.To evaluate these criteria, we used our own platform to build an investigator agent that reviews your criteria and evaluates performance much more effectively than traditional LLM-as-a-judge approaches.To get started visit dashboard.lucidic.ai and <a href="https:&#x2F;&#x2F;docs.lucidic.ai&#x2F;getting-started&#x2F;quickstart">https:&#x2F;&#x2F;docs.lucidic.ai&#x2F;getting-started&#x2F;quickstart</a>. You can use it for free for 1,000 event and step creations.Look forward to your thoughts! And don’t hesitate to reach out at team@lucidic.ai

Launch HN: Lucidic (YC W25) – Debug, test, and evaluate AI agents in production

Emacs: The macOS Bug

Hey HN,We are Winston, Edward, and James, and we built Meka Agent, an open-source framework that lets vision-based LLMs execute tasks directly on a computer, just like a person would.Backstory:In the last few months, we&#x27;ve been building computer-use agents that have been used by various teams for QA testing, but realized that the underlying browsing frameworks aren&#x27;t quite good enough yet.As such, we&#x27;ve been working on a browsing agent.We achieved 72.7% on WebArena compared to the previous state of the art set by OpenAI&#x27;s new ChatGPT agent at 65.4%. You can read more about it here: <a href="https:&#x2F;&#x2F;github.com&#x2F;trymeka&#x2F;webarena_evals">https:&#x2F;&#x2F;github.com&#x2F;trymeka&#x2F;webarena_evals</a>.Today, we are open sourcing Meka, our state of the art agent, to allow anyone to build their own powerful, vision-based agents from scratch. We provide the groundwork for the hard parts, so you don&#x27;t have to:* True vision-based control: Meka doesn&#x27;t just read HTML. It looks at the screen, identifies interactive elements, and decides where to click, type, and scroll.* Full computer access: It&#x27;s not sandboxed in a browser. Meka operates with OS-level controls, allowing it to handle system dialogues, file uploads, and other interactions that browser-only automation tools can&#x27;t.* Extensible by design: We&#x27;ve made it easy to plug in your own LLMs and computer providers.* State-of-the-art performance: 72.7% on WebArenaOur goal is to enable developers to create repeatable, robust tasks on any computer just by prompting an agent, without worrying about the implementation details.We’d love to get your feedback on how this tool could fit into your automation workflows. Try it out and let us know what you think.You can find the repo on GitHub and get started quickly with our hosted platform, <a href="https:&#x2F;&#x2F;app.withmeka.com&#x2F;" rel="nofollow">https:&#x2F;&#x2F;app.withmeka.com&#x2F;</a>.Thanks,
Winston, Edward, and James

Show HN: Open-source alternative to ChatGPT Agents for browsing

A short post on short trains

Fixing Ctrl+C in Rust terminal apps: Child process management