Cyrus
@cyrusnewday
Jhanas and AI. Cofounder @ZenbaseAI (YC S24), Core @ StanfordnNLP/DSPy | Past @CamelAIOrg @NousResearch @WomboAI @UWaterloo SYDE
Today we're launching Subconscious: a new platform for building agents with long-horizon reasoning and tool use, backed by MIT research. One API call. Tool use. Context beyond existing limits. If you're building agents, let's talk.
Getting all my wisdom teeth removed was one of my most ecstatic / psychedelic experiences Kinda odd I know but it’s amazing what happens when you don’t resist and dive headfirst into it
🤣 I think I style have some domain like arewedatingthesameguy.ai
The new #2 on App Store is a woman-only app that’s lets you review men you date! Tea Dating Advice replicates “Are we dating the same guy” FB groups. It also lets women reverse image search, background check and set alerts for names to catch cheaters. Also, the founder’s a guy.
Use Gemini CLI to capture your coding style by recursively analyzing your code :-)
🚀 Introducing context42 The best code style guide is the one your team already follows. This tool discovers it. context42 explores your codebase using Gemini CLI and conjures up custom style guides for every language it finds.
Lovable just raised $200M at a $1.8B valuation led by Accel. This all started unexpectedly with me calling my friend at 6AM to go for a walk. I've never shared this story before: (thread)
prompts as data structures feels very natural (can you spot the bug?)
woah claude code recovered from a bad architectural direction btw new n&w hack coming soon
I'm joining @eriktorenberg at @a16z on the new media team. Our mission: curate the most important conversations in technology (via podcasts, writing, and group chats), capture our world as it transforms, and write the history of the future in advance. If that compels you, DM me.
This is really cool work—markdown language for LLM prompting:
Just open-sourced LLML - think React but for AI prompts 🚀 Stop fighting with string concatenation. Start composing prompts from data structures. Available in Python, TypeScript, Rust (experimental), and Go (experimental)
1. Psilocybin helped human cells live longer in the lab. The cells stayed younger for longer and kept dividing more than usual before aging out. And the protective tips on DNA (telomeres), which wear down with cell divisions, stayed intact for longer. 2. Psilocybin…
Big news: we've figured out how to make a *universal* reward function that lets you apply RL to any agent with: - no labeled data - no hand-crafted reward functions - no human feedback! A 🧵 on RULER
# Why I'm deeply skeptical of prompt automation I believe good writing correlates with good thinking [1]. If you don't take the time to write a high quality prompt or coherent feedback about specific failures, the necessary human thinking never happens. You can hill…
on attachment to your current identity, and what it takes to evolve