Johannes Weniger

Automating AI research is the next major step in AI We let Claude Code (Opus 4.7) and Codex (GPT 5.5) run autonomously on the nanoGPT speedrun optimizer track using our idle compute. ~10k runs, ~14k H200 hours Opus now holds the record at 2930 steps vs the 2990 human baseline

🚆 One journey. One ticket. Full rights. Travelling by train across Europe can still be complicated, especially across borders and multiple providers. We want to change that. Our new package will make it easier ⬇️

@johannes_hage @burnpiro champ ! 🚴‍♀️ 👨‍💻 @burnpiro

bro took a week off from coding to compete in the 3,000+ km ultracycling world championship. truly built different @burnpiro

The next wave of AI will not be won by better prompts. It will be won by systems that learn from experience. Today, Prime Intellect Lab is out of beta, open for you to start training your own models. The era of self-improving agents is here.

Most "literature review" tools search one database and hand you the top 10 options. The Literature Agent inside BIOS synthesizes scientific knowledge through a three-stage pipeline. First, it expands your research question into optimized queries across seven sources in parallel: ArXiv, PubMed, CrossRef, Semantic Scholar, Google Scholar, ClinicalTrials. gov, and UniProt. Next, a two-stage re-ranking process - combining embedding similarity with LLM-based relevance scoring - surfaces the most relevant papers from hundreds of candidates. Two modes support different workflows. > Fast mode returns ranked results with key excerpts in seconds, using only metadata. > Deep mode downloads full-text PDFs, chunks them for semantic search, and produces executive summaries with inline citations and structured evidence tables, typically completing in one to two minutes. The result is a literature review that the agent has actually read.

you can just RL things

What BIOS can do in a single prompt: → "Search for biomarkers of X, then analyze my dataset for those specific markers" → "Find the standard analysis pipeline for this data type, then apply it to my data" → "Identify key genes from literature, then check their expression in my samples" BIOS orchestrates specialized subagents in parallel within each research cycle. Most powerful when combining literature search with data analysis.

Most AI tools that "do data analysis" write one block of code, run it, and call it done. The Data Analysis Agent inside BIOS works the way a careful analyst would. It breaks your task into smaller steps, writes Python code to execute each step, observes the results, and reflects on what it learned. The part most agents skip: a persistent knowledge base. Two kinds of memory get saved as it works: → Rules: extracted from documentation and domain conventions → Context: schema definitions, computed facts, and data quality caveats discovered during execution This structured memory ensures the agent doesn't repeat mistakes. Once the agent figures out something in your data, it carries that forward into the next step, the next iteration, the next session. The result is a multi-step analysis that builds on itself, using a memory of what it already learned. Try BIOS now: chat.bio.xyz

BIOS Dev Update: Fast Chat, Thinking Traces & More What's New: • Fast Chat Mode: A faster pipeline for simple questions that switches to literature search only when needed. It runs in an agentic loop, searching as many times as it needs until it has enough context to give you a complete answer. • Thinking Traces on Screen: BIOS now shows its reasoning on the main loading screen while Deep Research is running, so you can follow along in real time.

@dominik_scherm is this availalbe ? :)

built a small agent command center (acc) to keep track of all different pi, codex and claude sessions I have running

L’être humain est fascinant.

What does it actually look like to vibe code a dog medicine? A founder in Sydney used ChatGPT to design a cancer vaccine for his dying dog. @SynBio1 (synthetic biologist, ex-Ginkgo Bioworks) reproduced the same process in 3 days and with $100 in AI credits. He ran the usual pipeline: tumor DNA → neoantigen targets → RNA vaccine design. Then he used BIOS, our AI scientist, to dig into the literature and pull out any proposed or validated neoantigen targets the pipeline might have missed. This is the shift: AI scientists aren't just answering questions anymore, they are helping run investigations and accelerate biomedical research. ↓

The moment a research idea begins its transformation into intellectual property is less a single event than a long arc of infrastructure. Molecule is built around that entire journey, from a researcher's first onchain record of an idea, through datarooms, funding mechanisms, agentic tooling, real-world asset frameworks, community, and eventually IP licensing and spinout. We exist to serve every stage of that process. We've updated our brand to reflect something we've long understood: science and intellectual property aren't separate territories. They exist on a continuum, fluid and inseparable, held together by technical rigour. Vision directed by @SiMaclennan, who created the first Molecule logo in 2018.

DeSci Berlin is back June 18-19, during @BerBlockWeek This year's theme: Agentic Science & Scientific Singularity. What happens when AI starts doing science autonomously? Bringing together AI researchers, founders, and builders working at the frontier of autonomous agents, AI-driven drug discovery, consciousness research, and decentralized compute. Register to attend or apply to speak below ↓

what if dreaming is just auto context compaction for humans

automated ai researcher experiment (before autoresearch release) by geeve

@vincentweisser @msharmavikram @PrimeIntellect 🤌🏻👍

@msharmavikram @PrimeIntellect super excited to work more closely together!! 🙏

Incredible proud of @PrimeIntellect ❤️ You read it right - we now support RL to certain degree. Give it a try!

@afurgs hasta luego

hasta mañana

Excited to join NVIDIA GTC ahead of Jensen's keynote next Monday to talk about the agentic AI inflection point and building the self-improving agent stack Together with @steipete @SGRodriques @hwchase17, and moderated by @saranormous and @Alfred_Lin nvidia.com/gtc/pregame/