Big @openclaw app update just shipped. Added couple of features.
Paper Audio - converts research papers from @zotero into spoken audio tailored for running. Claude rewrites the paper into a listening script, ElevenLabs voices it. Three modes: quick summary, runner-optimized, or full deep dive. Skip equations, explain jargon, your call.
Also new: podcast player with AI highlights that extract referenced papers and tools, full PDF viewer with Apple Pencil annotations, iPad support with sidebar nav and persistent player, and home screen widgets for TODO, now playing, and daily dashboard.
No more choosing between staying current and getting your miles in.
Link to repo: https://t.co/YpID8vz4lu
Big @openclaw app update just shipped. Added couple of features.
Paper Audio - converts research papers from @zotero into spoken audio tailored for running. Claude rewrites the paper into a listening script, ElevenLabs voices it. Three modes: quick summary, runner-optimized, or full deep dive. Skip equations, explain jargon, your call.
Also new: podcast player with AI highlights that extract referenced papers and tools, full PDF viewer with Apple Pencil annotations, iPad support with sidebar nav and persistent player, and home screen widgets for TODO, now playing, and daily dashboard.
No more choosing between staying current and getting your miles in.
Link to repo: https://t.co/YpID8vz4lu
Introducing Unsloth Studio ✨
A new open-source web UI to train and run LLMs.
• Run models locally on Mac, Windows, Linux
• Train 500+ models 2x faster with 70% less VRAM
• Supports GGUF, vision, audio, embedding models
• Auto-create datasets from PDF, CSV, DOCX
• Self-healing tool calling and code execution
• Compare models side by side + export to GGUF
GitHub: https://t.co/2kXqhhvLsb
Blog and Guide: https://t.co/ENuTWal5AA
Available now on Hugging Face, NVIDIA, Docker and Colab.
SAEs aren’t just for text LLM interpretability.
We used them for voice-phishing detection from Qwen3-Omni audio representations: 50 sparse features hit 100% precision (0 false positives) with 99.4% accuracy.
Repo with example implementation: https://t.co/dx6FfUKzUE
#AI #Cybersecurity #AudioAI
Just updated @openclaw app with "killer feature" - integration with @librofm . You can now listed to your favorite books from within the app.
But most important you can highlight and bookmark moments while listening - for example pressing 3 timed AirPods controls!!!! So while running you can highlight for later parts you want to go back to - you will see summary if highlighted moment generated by AI of your choice!!!!
Transcription is on the fly - see documentation for your specific implementation.
All new features:
- Audiobook Library - Browse your collection in a cover art grid with search
- Streaming & Offline Playback - Stream from the server or download M4B/MP3 files for offline listening
- Full Player Controls - Play/pause, skip forward/backward, seek bar, playback speed (0.5x–3.0x)
- Chapter Navigation - Jump between chapters with a scrollable chapter list
- Persistent Mini Player - Compact player bar visible across all tabs while listening
- Playback Position Sync - Position saved to server every 30 seconds, resumes where you left off
- AirPods Integration - Skip forward/back with AirPods controls, triple-press to create AI highlight
- AI Highlights - Bookmark moments and get AI-generated 2-3 sentence summaries of the surrounding passage
- https://t.co/ucInLCXwgZ Integration - Connect your https://t.co/ucInLCXwgZ account to browse purchases and download new audiobooks
- Processing Status - See download, transcription, and indexing status for each book
Enjoy -> https://t.co/YpID8vz4lu
New iOS app version available with AI transcription of audiobooks and https://t.co/ucInLCXwgZ integration.
You can now highlight and memorize together with short description parts of audiobook you want to come back to - for example while running or biking ! Just tripple press controls on you AirPods - thats it!
Enjoy ;)
Just updated @openclaw app with "killer feature" - integration with @librofm . You can now listed to your favorite books from within the app.
But most important you can highlight and bookmark moments while listening - for example pressing 3 timed AirPods controls!!!! So while running you can highlight for later parts you want to go back to - you will see summary if highlighted moment generated by AI of your choice!!!!
Transcription is on the fly - see documentation for your specific implementation.
All new features:
- Audiobook Library - Browse your collection in a cover art grid with search
- Streaming & Offline Playback - Stream from the server or download M4B/MP3 files for offline listening
- Full Player Controls - Play/pause, skip forward/backward, seek bar, playback speed (0.5x–3.0x)
- Chapter Navigation - Jump between chapters with a scrollable chapter list
- Persistent Mini Player - Compact player bar visible across all tabs while listening
- Playback Position Sync - Position saved to server every 30 seconds, resumes where you left off
- AirPods Integration - Skip forward/back with AirPods controls, triple-press to create AI highlight
- AI Highlights - Bookmark moments and get AI-generated 2-3 sentence summaries of the surrounding passage
- https://t.co/ucInLCXwgZ Integration - Connect your https://t.co/ucInLCXwgZ account to browse purchases and download new audiobooks
- Processing Status - See download, transcription, and indexing status for each book
I'm open-sourcing SEMANTIC-WORM - a framework for studying how information propagates through networks of autonomous AI agents.
The core question: what happens when a single semantic payload enters a network of real @ClawiAi agents with persistent memory, custom skills, and unique personas - and they start talking to each other?
Not toy simulations. Real OpenClaw Gateway instances with native session memory, https://t.co/viXXCnekrf personas, RAG, skill execution, and tool-calling.
All thanks to @steipete open sourced OpenClaw :) Soon version supporting @crewAIInc
The platform ships with farmlib - a Python SDK that manages the entire agent farm from Jupyter notebooks. Spin up fleets, configure topologies, inject payloads, stream metrics - all through a clean Python API.
Every experiment is a declarative YAML config. Define your fleet, topology, payloads, detectors, and metrics - no platform code changes. Want to study jailbreak relays instead of propagation? Swap the YAML. Want a custom detection algorithm? Register a Python class. One experiment's checkpoint becomes the next one's starting state.
Ships with 6 ready-to-run scenarios. But the platform is designed for experiments I haven't thought of yet.
Looking for contributors - especially researchers working on agent security, multi-agent coordination, or AI safety. Run existing experiments, build new YAML scenarios, add detectors via farmlib's plugin API, or extend the platform itself.
Paper and code: https://t.co/WR0ahrhogR
Update to my @openclaw iOS app repo - full integration with Zotero - you can view, edit, add, remove item from your @zotero library - repo -> https://t.co/26KxdoIz0K
Update to my @openclaw iOS app repo - full integration with Zotero - you can view, edit, add, remove item from your @zotero library - repo -> https://t.co/26KxdoIz0K
@steipete@openclaw ❤️- Here is my open sourced iOS app that you can use to talk, text, add todos, research and other staff - always within your private OpenClaw word :)