Lance Skarada

need more hackathons at 5-star hotels

If you sleep 8 hours a night, you'll spend roughly 26 years of your life asleep. People with rare short-sleeper mutations (DEC2, ADRB1) need only 4-6 hours, with no health penalty. They get back 7-13 years of waking life. The molecule that controls this is orexin. Your brain's wakefulness switch. Block it, you sleep better (3 drugs already do this). Activate it, you need less sleep. Pharma is racing to build that activator: @EliLillyandCo paid $6.3B, @TakedaPharma expects @US_FDA approval this year. Now @paulkhls and @BioProtocol designed a selective orexin activator using AI in 24 hours, for $500 in lab costs. Instead of targeting narcolepsy like everyone else, they're going after ADHD. Nobody in the pipeline is doing that. Early stage, real obstacles (peptide half-life, brain delivery), but the speed of this is remarkable. @paulkhls curious about your delivery strategy for getting a peptide past the blood-brain barrier?

I packaged up the "autoresearch" project into a new self-contained minimal repo if people would like to play over the weekend. It's basically nanochat LLM training core stripped down to a single-GPU, one file version of ~630 lines of code, then: - the human iterates on the prompt (.md) - the AI agent iterates on the training code (.py) The goal is to engineer your agents to make the fastest research progress indefinitely and without any of your own involvement. In the image, every dot is a complete LLM training run that lasts exactly 5 minutes. The agent works in an autonomous loop on a git feature branch and accumulates git commits to the training script as it finds better settings (of lower validation loss by the end) of the neural network architecture, the optimizer, all the hyperparameters, etc. You can imagine comparing the research progress of different prompts, different agents, etc. github.com/karpathy/autor… Part code, part sci-fi, and a pinch of psychosis :)

A randomized trial of using an LLM by primary care physicians for referral to specialists (vs no AI) provided substantial improvements in workflow, patient experience and less test ordering @NatureMedicine nature.com/articles/s4159…

New research from Google DeepMind. Really good read to understand what's ahead with AI agents. It introduces a comprehensive framework for intelligent AI delegation, a sequence of decisions involving task allocation that also incorporates transfer of authority, responsibility, accountability, clear role specifications, and mechanisms for establishing trust between parties. Why does it matter? As AI agents move from isolated assistants to participants in complex delegation networks and virtual economies, the absence of robust delegation protocols introduces significant societal risks. This framework aims to inform the development of protocols for the emerging agentic web. Paper: arxiv.org/abs/2602.11865 Learn to build effective AI agents in our academy: academy.dair.ai

Vector databases just got disrupted 🤯 You can now build RAG without Vector DBs. PageIndex is a new open-source library that uses document trees instead of embeddings. It achieves 98.7% on FinanceBench by letting LLMs reason over structure rather than matching keywords. → No Embeddings → No Chunking 100% Open Source.

1/🧵How do we know if AI is actually ready for healthcare? We built a benchmark, MedHELM, that tests LMs on real clinical tasks instead of just medical exams. #AIinHealthcare Blog, GitHub, and link to leaderboard in thread!

@ShramanKar Super excited for this

Today, we are launching Golpo, the world’s most powerful AI video generation model to help you create AI-generated explainer and animation videos. With just a prompt, Golpo turns your ideas into stunning animated and whiteboard-style videos Enjoy P.S. If you want some free credits, please drop your email in the comments below. Here’s a Golpo… explaining Golpo

Here's what we created at the Coinbase AI Hackathon! Agent Access is a decentralized marketplace for AI Agents, powered by @CoinbaseDev and @MorpheusAIs Our flagship agent, sAIgent, is a swarm of sentiment analysis agents which researches specific token addresses. @mahi_jariwala @0xyoshii @domsteil @DJohnstonEC @virtuals_io @base @kleffew94 @nemild @NEARProtocol @hyperbolic_labs @openmind_agi @coinbase

Had a blast at the Coinbase AI Hackathon! Met the best people and learned so much. Building a subnet on chain and ai agents marketplace in hours was not on my 2025 checklist. @MurrLincoln @nemild @kleffew94 @DJohnstonEC @domsteil @mahi_jariwala @MorpheusAIs @base @coinbase

@JackedApeClub Me

Who still doesn't have a whitelist? Reply to this tweet 💪🏼🐵

@EvaK360Boss @TastyBonesNFT @BRVHMESSI @LJSk45

Ill be giving out 3 #bonemarrows to @TastyBonesNFT in 24 hrs 🎉 To enter: - Follow @EvaK360Boss and @TastyBonesNFT - Like + RT - Tag some frens who should join The Land of the Living

@HAPEBEASTGANG @sophiemaialee @SereSalome @BRVHMESSI

Heating up to mint on the 19th with @sophiemaialee & @SereSalome 14/1/2022. Be there. 🦍🤍🍌™️