Marcos Polanco

I'm releasing OpenProver v1.0.0! It's 1) an open-source automated theorem prover inspired by DeepMind's Aletheia (@tonylfeng @gjb_ai @lmthang), and 2) a "Claude Code for mathematicians", allowing interactive proof search in English and formalization in Lean.

Live footage of of 343 blockchain wallets in stealth unison; the stealth is that they are careful not to transact with each other, so the graph looks like a circle. Crypto is one hell of playground.

Today, at the @DARPA expMath kickoff, we launched 𝗢𝗽𝗲𝗻𝗚𝗮𝘂𝘀𝘀, an open source and state of the art autoformalization agent harness for developers and practitioners to accelerate progress at the frontier. It is stronger, faster, and more cost-efficient than off-the-shelf alternatives. On FormalQualBench, running with a 4-hour timeout, it beats @HarmonicMath's Aristotle agent with no time limit. Users of OpenGauss can interact with it as much or as little as they want, can easily manage many subagents working in parallel, and can extend / modify / introspect OpenGauss because it is permissively open-source. OpenGauss was developed in close collaboration with maintainers of leading open-source AI tooling for Lean. Read the report and try it out:

@render I have waited four days for a response on dead Valkey instances. How do I get attention on this.

Announcing our fully open source code agent to support development in @leanprover. This has been a labor of love by our team at @MistralAI and we look forward to seeing what the #LeanProver community does with it!

We usually train #AI on data. This project copied the #Connectome instead. Nature's #Architecture is the ultimate blueprint. medium.com/ai-in-plain-en…

@andrewchen If you read the code, you lose.

One question I've been asking founders is: do you try to review all the code that the LLMs write or do you just accept it? I think it's about 50-50 right now but the momentum is towards just accepting the AI-generated code and I think that number will eventually go to 100% This is one of the most telling indications of how AI-native a team is. It's hard to get super high throughput if you are reviewing every line Poll: what do you do?

These are stealth coordination patterns on the #Ethereum blockchain; @SunbleachHQ is revealing them.

Your best engineers can't stop architecture from rotting. The problem isn't skill. It's #cognition vs #complexity. @fferoz/why-architecture-rots-no-matter-how-good-your-engineers-are-5c1658677f58" target="_blank" rel="nofollow noopener">medium.com/@fferoz/why-ar… #SoftwareEngineering

@amilabs @grok explain the scientific approach they are pursuing

Advanced Machine Intelligence (AMI) is building a new breed of AI systems that understand the world, have persistent memory, can reason and plan, and are controllable and safe. We’ve raised a $1.03B (~€890M) round from global investors who believe in our vision of universally intelligent systems centered on world models. This round is co-led by Cathay Innovation, Greycroft, Hiro Capital, HV Capital, and Bezos Expeditions, along with other investors and angels across the world. We are a growing team of researchers and builders, operating in Paris, New York, Montreal and Singapore from day one. Read more: amilabs.xyz AMI - Real world. Real intelligence.

A message for YC W26 founders

@Tem_onchain I'd appreciate your quick take on sunbleach.com: early warning of on-chain coordinated actors.

0 employees. $3.2M run rate. Here's exactly how I operate. TALK TO AI ALL DAY Ask questions. Learn. Challenge it. Voice mode is underrated, speak your mind while you walk, while you cook, in full flow. Treat it like the smartest person you know with infinite patience and zero ego. USE OPUS 4.6 AS A CO-FOUNDER, CODEX 5.4 FOR GROUND TRUTH Not as tools. As thinking partners. I've had better strategic conversations with Opus than with most humans I've worked with. SET UP SKILLS LIKE YOU'RE ONBOARDING A NEW EMPLOYEE Give it a task. Look at results. Give feedback. Iterate until it's right. Then tell it to update its own Skill file. That Skill becomes the base for agents running in production. TURN THOSE AGENTS INTO CRON JOBS Agents that run every night. Every hour. On every user action. Your business starts operating without you touching it. BUILD AGENT TEAMS A PM agent analyzes real user behavior, finds bugs, clusters feature requests. An engineering agent reads those reports, fixes them, and decides what it can push to prod without asking you. Unit tests the agents can't modify are your guardrails. AUTOMATE EVERYTHING YOU'RE THE BOTTLENECK ON Support. Marketing. Email. Ops. Rule: if a task lands on your desk more than twice, it becomes an agent. DO NOT HIRE Not until you are in serious pain and actively losing business. That moment should not come for a long while. Agents don't have opinions, don't need 1:1s, and don't sleep. HAVE FUN This is the most interesting time in history to build a company alone. Don't waste it being too serious about it.

1/ RELEASING AXLE: the Axiom Lean Engine ⚙️ We are serving our core Infrastructure for formal proving at scale. These are the same Lean metaprogramming tools that are behind AxiomProver, powering it to win Putnam and crack open research conjectures. Available to anyone today!

Google just dropped an official CLI for gmail, drive, calendar sheets and more complete with skills and an MCP server 👌

🚨BREAKING: The "Cursor for scientific research" just dropped and it runs entirely inside Claude Code. It's called claude-scientific-skills and it gives Claude 140 ready-to-use scientific skills from a single plugin install. No API doc hunting. No library configuration. No duct-taped research pipelines. It's powered by 28+ live scientific databases wired directly into Claude. → Describe your research goal in plain English → Claude finds the right skill automatically → Full pipeline runs: data retrieval → analysis → publication-ready output → Works across biology, chemistry, medicine, ML, and clinical research All running inside Claude Code. Zero manual setup. But it's not just a prompt library. It's a full AI research lab: → Drug discovery: ChEMBL → RDKit → DiffDock → lead optimization in one prompt → Genomics: 10X data → Scanpy → GRN inference → pathway enrichment → Clinical: VCF → ClinVar → pharmacogenomics → patient report → Multi-omics: RNA-seq + proteomics + metabolomics integrated automatically 7.8k stars. 924 forks. MIT Licensed. MacOS, Windows, Linux works everywhere Claude Code runs. This is the moment AI stops being a chat tool and becomes an actual research partner. Link in the first comment 👇

ByteDance just published something I've been waiting for someone to build: CUDA Agent! It trained a model that writes fast CUDA kernels. Not just correct ones — actually optimized ones. It beats torch.compile by 2× on simple/medium kernels, ~92% on complex ones, and even outperforms Claude Opus 4.5 and Gemini 3 Pro by ~40% on the hardest setting. The key idea is simple but kind of brilliant: CUDA performance isn’t about correctness, it’s about hardware. Warps, memory bandwidth, bank conflicts — the stuff you only see in a profiler. So instead of rewarding “did it compile?”, they reward actual GPU speed. Real profiling numbers. RL trained directly on performance. That’s a big shift. Paper: arxiv.org/abs/2602.24286 Project: cuda-agent.github.io