Maya

@braza_faouzi @KexinHuang5 @ScienceMagazine You can do that and WAY more on @k_dense_ai !!

@KexinHuang5 @ScienceMagazine any chance to have a draft pdf ?

Today, we're excited to share that Biomni is published in @ScienceMagazine. Biomedical research is still fragmented, manual, and difficult to scale. In this work, we introduce Biomni - the first general-purpose biomedical AI agent with an integrated biology environment that can reason, plan, and execute end-to-end scientific workflows. We show that, with the right environment and harness, AI can automate large-scale omics analyses, orchestrate laboratory robotics, optimize molecular properties, and even train new AI models for biology. We also introduce a reinforcement learning recipe for continually improving biomedical AI agents, enabling open-source models to achieve frontier-level performance. It's surreal to look back. We started the Biomni project in early 2024, when agentic AI was still nascent. It is exciting to see tens of thousands of biologists collaborating with agents every day to accelerate science. Try Biomni: biomni.phylo.bio Read more: science.org/doi/10.1126/sc… This work is not possible without this truly inter-disciplinary team: @serena2z @hcwww_ @YuanhaoQ Minta Lu, Ryan Li, @yusufroohani Lin Qiu @shiyi_c98 Gavin Junze Di @rickwierenga @kavi_deniz Sherry @TianweiShe Shruti Jennefer Xin Zhou @MWheelerMD Jon Bernstein @MengdiWang10 @PengHeAtlas @zhou_jingtian @SnyderShot @lecong Aviv Regev @jure @StanfordAILab @genentech @phylo_bio @arcinstitute @UW @berkeley_ai @RetroBio_ @tamarindbio @Princeton @UCSF

@Shoier__ @KexinHuang5 @ScienceMagazine While you’re giving it a try, give @k_dense_ai a shot!

@KexinHuang5 @ScienceMagazine I tried it. Decent so far. Will report what I find later.

POV: you’re using @k_dense_ai BYOK as your research assistant 🧬 In this vlog, I walk through how I'm using BYOK — from organizing messy inputs to turning scattered notes and data into a more structured starting point for analysis. Bring your own API key. Bring your own research question. Let K-Dense help with the rest. Download BYOK today at github.com/K-Dense-AI/k-d…

Claude Science but with any model you like including local models. In use already by thousands of scientists worldwide. Please repost. github.com/K-Dense-AI/k-d…

Can’t wait to join this webinar and learn the basics myself!! 🤫

@jocarrasqueira @NanoBanana Yes we did!! You can read all about Nano Banana 2 Lite’s ability to generate scientific images over at our @k_dense_ai blog: k-dense.ai/blog/benchmark…

And you, have you tried the new @NanoBanana 2 Lite model? 🍌

@osanseviero We @k_dense_ai got the early access scoop to benchmark this model for scientific images! For those interested 👀👀 k-dense.ai/blog/benchmark…

Introducing Nano Banana 2 Lite 🍌⚡️Faster, cheaper, and better than Nano Banana 1 I've been having tons of fun over the last couple of weeks playing with this model. The quality is really impressive for its speed. I'm looking forward to seeing what you all build with it!

@vamsibatchuk @GoogleDeepMind We @k_dense_ai had early access to the model and LOVED the speed!! You can read all about our insights from the early access here: k-dense.ai/blog/benchmark…

My first launch with @GoogleDeepMind It was incredibly fun to work on this project !! Nano Banana 2 Lite is here, everyone 🍌⚡️ insanely fast !! try it out and share examples

@AllThingsApx All the more reason why early access to these platforms is so important! We @k_dense_ai tested the #BioNeMo agent toolkit integrated with our skills… read more here!: k-dense.ai/blog/benchmark…

The Life Sciences are the hardest proving ground for AI. Excited to see Anthropic double down with Claude Science and #BioNeMo Agent Toolkit today. anthropic.com/news/claude-sc…

@rand_longevity 👀👀 Sounds exactly like the kind of agentic skills we’ve been developing over @k_dense_ai!! Check out the skills for yourself… github.com/K-Dense-AI/k-d…

this is way better than sonnet, forget chatbots I want agents running science

@mark_k @GoogleDeepMind +1 to your testing!! Our team @k_dense_ai had early access to the Nano Banana 2 Lite model and tested it for scientific image generation 👀👀 You can read more here on our blog: k-dense.ai/blog/benchmark…

I tested Nano Banana 2 Lite by @GoogleDeepMind so you don't have to! 🍌 Here is a complex image I created with the model for testing. * It's extremely fast (about 3s) * Text rendering is so-so. Smaller text can be garbled. * Overall it's quite impressive for the speed & cost!

@darekgusto Intriguing that you point out using the Nano Banana 2 Lite model for simple images. Our team @k_dense_ai had early access and found quality concerns too 👀👀 read more on our blog: k-dense.ai/blog/benchmark…

First impression? Slightly disappointed. New image model from Google / Gemini turned out a Nano banana 2 Lite variant. Sure, cost and speed is nice, and the default output is decent, but will I use it? Probably only for less complex illustrations. It would be different case if we could use Lite to spam images until I find a good base to refine with full Nano Banana 2. But with how Nano Banana is so terrible at keeping quality of edited images, it's a lost case. Even with Nano Banana 2 generated images, I'd rather go out of my way to edit them using Grok Imagine, because at least the quality won't drop so much. Will I still post some images and comparisons between variants? Possibly so. :)

@SadAlbert10 That’s what we found & more during @k_dense_ai’s early access to the Nano Banana 2 Lite version when looking at scientific image generation! Read more about it here: k-dense.ai/blog/benchmark…

It also Generates a lot faster right?😳 Right?😳😳

@dr_alphalyrae Try K-Dense BYOK - it goes further with a fully local, open desktop solution powered by our Scientific Agent Skills library (github.com/K-Dense-AI/sci…). It gives you 140+ expert skills, access to 100+ databases, and complete model freedom.

So basically stripped down version of our K-Dense BYOK but tied to Anthropic models. github.com/K-Dense-AI/k-d…

@IterIntellectus Nope. @k_dense_ai is more alive than ever!

@owenthcarey @claudeai You can already get a full brief on this pressing matter and other hot-topic debates (what is the best kind of coffee strain? pineapple on pizza?) using K-dense BYOK agentic skills 👀👀 github.com/K-Dense-AI/k-d…

@claudeai Me connecting to 60+ scientific databases just to research if a hot dog is a sandwich.

Introducing Claude Science, a new app designed with every stage of research in mind. Artifacts traced to their code, environments managed on demand, and 60+ optional scientific databases that you can connect. Available now in beta.

@claudeai Great to see more tools for AI + science! For anyone looking for an open-source, model-agnostic option, we’ve been building K-dense BYOK: github.com/K-Dense-AI/k-d…

Excited to share new work with @NVIDIAAI: we benchmarked 10 BioNeMo NIM skills across three @AnthropicAI Claude models, ~830 controlled runs. The result: skills don't make the models smarter, they make delivery reliable. On hard calls, a model ~5x cheaper became more reliable than the frontier baseline. k-dense.ai/blog/benchmark…