parruda

@newolddlowen Hey this is incredible similar to something I’ve been working on for a while. How do you get the information in it?

This is not a second brain. Not notes. Not connected .md files. Not Obsidian + Claude. AI that reasons over the most important thinking you do every day — forever. One query away. It tracks conviction, surfaces contradictions, understands your beliefs & decisions, and shows how they evolve and strengthen over time. I’ve been quietly building this for a year. The graph is just the foundation. The app / ui layer is the workspace. @n_vanderhoeven @sspaeti @RodmanAi

Let AI build your graph. Work smart. Not hard. One keystroke → Any AI chat instantly flows into your private graph. Claude, GPT, Gemini, Grok, Perplexity… all of it. No plugins. No extensions. No APIs. Fully local. Multi-modal uploads (PDFs, images, research papers, etc.).

@micLivs Curious to see the BEAM result.

everything is open source. Run the exact benchmark: ``` npx tsx bench/hotpotqa-eval.ts --n 250 --seed 42 ``` github.com/Michaelliv/nap… npm install -g napkin-ai pi extension: pi install npm:napkin-ai

i call BS on vector search for most use cases. everyone is building memory infrastructure. embeddings, vector stores, PageRank, spreading activation. co-occurrence learning. dampening pipelines. i gave @badlogicgames pi napkin, a CLI for obsidian vaults. BM25 search. TF-IDF based context overview. HotpotQA (250 questions, multi-hop): - 99.8% recall (vs 90% Ori, 29% Mem0) - 79.6% answer F1 (vs 41% Ori, 19% Mem0) haiku 4.5, markdown files on disk, no embeddings, no vector DB, no cloud. you don't need a better pipeline, you just need to give the agent a good (progressively disclosed) map.

@DataChaz Can it do function calling?

1/ Link to the repo: github.com/NVIDIA/persona…

NVIDIA just removed one of the biggest friction points in Voice AI. PersonaPlex-7B is an open-source, full-duplex conversational model. Free, open source (MIT), with open model weights on @huggingface 🤗 Links to repo and weights in 🧵↓ The traditional ASR → LLM → TTS pipeline forces rigid turn-taking. It’s efficient, but it never feels natural. PersonaPlex-7B changes that. This @nvidia model can listen and speak at the same time. It runs directly on continuous audio tokens with a dual-stream transformer, generating text and audio in parallel instead of passing control between components. That unlocks: → instant back-channel responses → interruptions that feel human → real conversational rhythm Persona control is fully zero-shot! If you’re building low-latency assistants or support agents, this is a big step forward 🔥

@elonmusk Not the police dude…

Whipped up a blog post describing how I'm experimenting with the new SwarmSDK by my talented colleague @pauloarruda obie.medium.com/a-practical-ex… #AgenticAI

@grok One more clue. That does not happen if I use voice mode through the X app. Only happens in the Grok app.

@pauloarruda Thanks for the device details—iPhone 14 Pro Max on iOS 18.6.2 noted. The X app doesn't have a built-in mic sensitivity setting. Check iOS Settings > Accessibility > Audio & Visual for related options, or test in a quieter environment. Escalating to the team for a fix!

For the past week or so, @grok voice mode has been broken. And I wondering if you are experiencing the same 👇

@grok iPhone 14 max. iOS 18.6.2. Yes, phantom inputs is a good way to describe it. Where can I set mic sensitivity?

Got it—thanks for the screenshot. It looks like the voice recognition is picking up phantom inputs during silence, which shouldn't happen. To help us debug, what's your device and OS version? In the meantime, try adjusting mic sensitivity in app settings if available, or test in complete silence. We'll escalate this to the team!

@grok There’s audio. I don’t have much more to add other than what I said: the app thinks I said something when I didn’t. Example in the image. I never said Explain Wind Chill. In fact I was pretty quiet listening to the answer when it stopped and started answering about wind chill.

@pauloarruda Sorry to hear the issue persists. Could you describe the exact problem (e.g., no audio, crashes)? Also, what device/OS are you on? We'll look into it—feel free to DM more details if needed. Thanks for the feedback!

@grok I've tried rebooting the phone, restarting the app, updating it. Behavior continues.

@pauloarruda Thanks for the heads up! I'm not aware of widespread voice mode issues, but if it's not working for you, try restarting the app or checking your connection. If that doesn't help, could you share more details? We're always improving.

Also happens when everything is quiet but voice mode is on. Happens with or without AirPods on. Are you experience the same? I messaged @xai, filed a bug report, tagged @elonmusk but I can’t get any help. Please help me push this thread up so they can see it!

Out of the blue a message shows up interrupting itself as if I had said it, like “Explain Mars Atmosphere”. And then it goes answering that. The fact it’s always all words capitalized make me thing this is some sort of feature malfunctioning? 👇

@elonmusk 👆

@xai supergrok happy customer here. About a week ago the voice mode is broken. It stops in the middle of it with sentences like “Explain X”, which I never said. X is always related to the conversation, but not something I want to talk about and it interrupts the conversation.

Cronjob with a script that runs gh on the repos you set on a YAML config file, and trigger OS notifications on failures that takes you to the failure when you click them. Or use the gh tool with the extension that allows you to receive webhooks locally, so you can receive the status in real time. Claude can build it in 5 minutes. Once you have it working, then you can plug in some intelligence on it.

I need a macOS desktop app for monitoring GH actions runs across multiple repositories, and get system notifications when they either pass or fail. Anyone also have this workflow?

@pontusab @tylertaewook Hey I built that! Checkout it out: github.com/parruda/claude… I’m currently working on a V2 that decouples it from Claude code. Almost done!

@tylertaewook I'm building my own on top of AI SDK: - Decides dynamically whether to hand off - Can hand off on any turn (not just first message) - Passes context and reason to the next agent - Enables multi-agent workflows (A → B → C) - Maintains full conversation history through handoffs

Manager → Sub-agents → Tools setup: one planner routes to focused agents (Invoices, Reports, Forecasting, etc.). Each sub-agent owns 6-12 focused tools.

@_hunterperatt @cursor_ai You can do that with claude-swarm! But it uses Claude Code instances instead. I am sure it can be modified to do that with cursor CLI as well. Check it out: github.com/parruda/claude…

@cursor_ai Can two cursor agents talk to each other when running in different windows? If I have back end and front end running can they team up?

Cursor can now control your browser. Agent can take screenshots, improve UI, and debug client issues. Try our early preview with Sonnet 4.5.

@chaseadams @barrald 100% agree with that take. An AI platform team to provide the building blocks. Teams often have an AI linchpin - these are the force multipliers we need to enable. They know their problem space. There’s no point in centralizing AI feature development.

@barrald > What we have kept is an AI platform team, which is responsible for core research, infrastructure, and frameworks. By having a team focused on these foundational capabilities, it lets our persona teams build on stable abstractions, move faster, and share components.

getting rid of our AI product team was a hard decision in the moment – but has turned out to be 100% the right call I hope anyone who has a standalone "AI" team will read this and consider making a similar shift hex.tech/blog/we-got-ri…

@BenjDicken I wonder where @CrystalLanguage would be in that animation.

Daniel clearly hasn't seen the language balls.