Daniel Burkhardt

Digital Biology @ NVIDIA

[Daniel Burkhardt @DBBurkhardt]

@AndrewLBeam @jonathanbeam @reflection_ai This is why I open twitter

[Daniel Burkhardt @DBBurkhardt]

At NVIDIA, we have the privilege of working with companies that are changing how care is delivered. We’re an accelerated computing company, but our partners bring us into problems I might never learn about otherwise.

For Breast Cancer Awareness Month, @NVIDIA Inception is highlighting four companies working on different parts of breast cancer care.

At @isonohealth, researchers are working to solve for the variation in handheld ultrasound between different operator’s technique, which makes comparing scans across visits difficult. Their automated system captures a standardized scan in about two minutes, giving clinicians a more repeatable view to track changes.

@whiterabbitai is tackling the workload of breast cancer screening, where radiologists review large numbers of mammograms to find relatively few cancers. Alongside its breast-density assessment software, the team is researching how AI could identify negative scans automatically and help radiologists focus on cases needing closer review.

Then there’s the question of what to do after a diagnosis. @Ataraxis_AI uses pathology slides and clinical data already collected during care to help assess recurrence risk and whether chemotherapy is likely to benefit a patient. This means getting more from existing data and reducing extra scans.

@SimBioSys focuses on treatment planning. Surgeons need to understand how a tumor sits relative to the tissue around it. The company builds 3D models that make those relationships easier to visualize and help inform surgical decisions.

More on each of these companies and how NVIDIA is accelerating their work here: https://t.co/c1gj1MjuBk

[Daniel Burkhardt @DBBurkhardt]

How is AI accelerating scientific progress?

On October 7, I'm moderating a conversation with three founders and leaders working on that question:
• Michelle Lee, CEO of @medra_ai
• Andy Beam, CTO of @LilaSciences
• Gleb Kuznetsov, CEO of @ManifoldBio

@AnthropicAI and @nvidia are bringing together founders, researchers, and builders at Claude Founder House during #SFTechWeek to talk about AI for science and drug discovery.

We’ll start with a short panel, then continue the conversation over drinks. If you’re building in this space, we’d love to see you there.

October 7 · 5–7:30pm PDT · San Francisco

Request to join: https://t.co/gJDVK5CT7N

[Daniel Burkhardt @DBBurkhardt]

@AlexanderTong7 @schmidtsciences Way to go man, you deserve it!

[Daniel Burkhardt @DBBurkhardt]

One of my favorite things about hackathons is seeing what happens when you give people new tools and let them just build.

We spent three days in London with @OpenAI, @NVIDIAHealth, @WilbeLABS, and 75 scientists and engineers exploring ChatGPT, Rosalind Workbench, NVIDIA BioNeMo, and agentic AI.

The projects went in completely different directions—from systems to make agentic workflows reproducible, to screening herbicide-resistant plant proteins, to identifying chimeric RNAs as potential drug targets.

One of the biggest takeaways for me was seeing the boundary between scientific and engineering work start to blur. Scientists can use agents to turn ideas into working code much more directly. Engineers can use them to get closer to the domain knowledge behind the workflows they're building.

Both skillsets are critical for the future of discovery. Agents are helping bridge the gap to make it much easier for scientists and engineers to build together.

It's a pretty exciting glimpse of the future.

[Daniel Burkhardt @DBBurkhardt]

It's been amazing working with @joyjiao12 and the OpenAI Life Sciences team to bring accelerated computing and frontier intelligence to advance life sciences research.

We've been working with GPT-Rosalind and the Rosalind Workbench to explore how developers will use these tools with @nvidia BioNeMo Agent Toolkit for genomic analysis, protein binder design, molecular optimization, and more.

I can echo @ChrisHayduk, it feels like living in the future.

[Chris Hayduk @ChrisHayduk]

At OpenAI, it's felt like our researchers & engineers have been living in the future with Codex.

Today, we're bringing that same experience to life science researchers with the Rosalind Workbench.

[Chris Hayduk @ChrisHayduk]

At OpenAI, it's felt like our researchers & engineers have been living in the future with Codex.

Today, we're bringing that same experience to life science researchers with the Rosalind Workbench.

[NVIDIA Healthcare @NVIDIAHealth]

🔬 Scientific AI should help researchers move faster - without taking them further from the evidence.

We’re excited to support the Rosalind Workbench research preview with the NVIDIA BioNeMo Agent Toolkit, bringing agent-ready capabilities for genomics, molecular structure prediction, biological sequence analysis, and more directly into the workspace.

AI can help researchers plan and execute scientific tasks - while people remain in control of the review, interpretation, and validation that make science trustworthy.

https://t.co/lxTIFkaGp8

[Daniel Burkhardt @DBBurkhardt]

Very cool seeing Biomni by @KexinHuang5 and team published in Science! Congrats 🎉

Looking forward to seeing more updates from @phylo_bio as they continue to build out AI tools for scientists

[Kexin Huang @KexinHuang5]

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: https://t.co/Pzm8TVlkNA
Read more: https://t.co/pRstxzPKDi

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

[Daniel Burkhardt @DBBurkhardt]

What I find most interesting about agents is how much freedom they have inside a fixed compute budget. Give an agent a goal and it can write code, search the web, and operate a computer in ways you never prescribed. The moment it became real for me was using coding agents to build a macOS transcription app in Swift, a language I had never used.

That left me wondering, which problems benefit most from giving agents this much freedom?

I think recursive self-improvement sits at the far end of this spectrum, where agents edit and optimize their own code. It works well when the task has a deterministic verifier, like software engineering and unit tests. It gets harder for open-ended tasks like writing papers or natural-language proofs, where evaluation relies on model-based judges. Agents can exploit weaknesses in those judges, creating an alignment gap between what earns a high score and what people actually value.

For the past three months, I’ve had the privilege of working on this problem with @Alex__Iacob, @itsmaddox_j, @williamfshen, @niclane7, and a team spanning @Cambridge_Uni, @NVIDIAHealth, @flwrlabs, @mbzuai, and @Inria.

We recently released the Red Queen Gödel Machine, which co-evolves agents with their evaluators. It reached a 71.7% coding pass rate versus 69.9% for the prior baseline using 1.35–1.72× fewer tokens, produced paper writers accepted 1.78–1.86× more often by a reviewer panel, and improved proof-grading accuracy from 70% to 76%.

I also led experiments with @NickVenanzi using Nemotron 3 Ultra for search-time worker agents and GPT-5.5 for the meta-agent. On paper review, this approached the GPT-5.5-only result at roughly 13× lower price-equivalent search-token cost.

Read the paper: https://t.co/JnkrxaKzl4

[Daniel Burkhardt @DBBurkhardt]

@anshulkundaje @CalebLareau I heard labs pay extra for training tokens from @CalebLareau

[Theodoris Lab @TheodorisLab]

Excited to share MaxToki, a temporal AI model that predicts the impact of perturbations on cell states over time along dynamic trajectories. We applied MaxToki to predict how cells age across the human lifespan & discovered new cardiac pro-aging drivers that we validated in vivo.

[Edison Scientific, Inc @EdisonSci]

At Edison Scientific, we build agents for scientific discovery. Training these agents often requires scaling to 1000s of simultaneous environments, for which we rely on @nvidia's recently announced NeMo Gym and RL frameworks. Read more about the integration with Aviary

[NVIDIA Healthcare @NVIDIAHealth]

🤯 Discover how you can scale transformer models to billions of parameters in PyTorch - without rewriting your pipeline.

Dive into NVIDIA BioNeMo Recipes to see how FP8 acceleration, efficient sequence packing, and the Transformer Engine empower breakthrough R&D on NVIDIA GPUs.

Ready to transform model training?
Read the blog 👉 https://t.co/vVuLbFjOzO

[Sam Rodriques @SGRodriques]

Today, we’re announcing Kosmos, our newest AI Scientist, available to use now.

Users estimate Kosmos does 6 months of work in a single day. One run can read 1,500 papers and write 42,000 lines of code. At least 79% of its findings are reproducible. Kosmos has made 7 discoveries so far, which we are releasing today, in areas ranging from neuroscience to material science and clinical genetics, in collaboration with our academic beta testers. Three of these discoveries reproduced unpublished findings; four are net new, validated contributions to the scientific literature. AI-accelerated science is here.

Our core innovation in Kosmos is the use of a structured, continuously-updated world model. As described in our technical report, Kosmos’ world model allows it to process orders of magnitude more information than could fit into the context of even the longest-context language models, allowing it to synthesize more information and pursue coherent goals over longer time horizons than Robin or any of our other prior agents. In this respect, we believe Kosmos is the most compute-intensive language agent released so far in any field, and by far the most capable AI Scientist available today. The use of a persistent world model also enables single Kosmos trajectories to produce highly complex outputs that require multiple significant logical leaps. As with all of our systems, Kosmos is designed with transparency and verifiability in mind: every conclusion in a Kosmos report can be traced through our platform to the specific lines of code or the specific passages in the scientific literature that inspired it, ensuring that Kosmos’ findings are fully auditable at all times.

We are also using this opportunity to announce the launch of Edison Scientific, a new commercial spinout of FutureHouse, which will be focused on commercializing our agents and applying them to automate scientific research in drug discovery and beyond. Edison will be taking over management of the FutureHouse platform, where you can access Kosmos alongside our Literature, Molecules, and Precedent agents (previously Crow, Phoenix, and Owl). Edison will continue to offer free tier usage for casual users and academics, while also offering higher rate limits and additional features for users who need them. You can read more about this spinout on our blog, below.

A few important notes if you’re going to try Kosmos. Firstly, Kosmos is different from many other AI tools you might have played with, including our other agents. It is more similar to a Deep Research tool than it is to a chatbot: it takes some time to figure out how to prompt it effectively, and we have tried to include guidelines on this to help (see below). It costs $200/run right now (200 credits per run, and $1/credit), with some free tier usage for academics. This is heavily discounted; people who sign up for Founding Subscriptions now can lock in the $1/credit price indefinitely, but the price ultimately will probably be higher. Again, this is less chatbot and more research tool, something you run on high-value targets as needed.

Some caveats are also warranted. Firstly, we find that 80% of Kosmos findings are reproducible, which also means 20% are not -- some things it says will be wrong. Also, Kosmos certainly does produce outputs that are the equivalent to several months of human labor, but it also often goes down rabbit holes or chases statistically significant yet scientifically irrelevant findings. We often run Kosmos multiple times on the same objective in order to sample the various research avenues it can take. There are still a bunch of rough edges on the UI and such, which we are working on. Finally, we are aware that the 6 month figure is much greater than estimates by other AI labs, like METR, about the length of tasks that AI Agents can currently perform. You can read discussion about this in our blog post.

Huge congratulations to our team that put this together, led by @ludomitch and @michaelathinks: Angela Yiu, @benjamin0chang, @sidn137, Edwin Melville-Green, Albert Bou, @arvissulovari, Oz Wassie, @jonmlaurent. A particular shout out to @m_skarlinski and his team that rebuilt the platform for this launch, especially Andy Cai @notAndyCai, Richard Magness, Remo Storni, Tyler Nadolski @_tnadolski, Mayk Caldas @maykcaldas, Sam Cox @samcox822 and more.

This work would not have been possible without significant contributions from academic collaborators @mathieubourdenx, @EricLandsness, @bdanubius, @physicistnevans, Tonio Buonassisi, @BGomes_1905, Shriya Reddy, @marthafoiani, and @RandallBateman3.

We also want to thank our numerous supporters, especially @ericschmidt, who has been a tremendous ally. We will have more to say about our supporters soon!

[Edison Scientific, Inc @EdisonSci]

Edison Scientific is launching today, with our launch of Kosmos, the most powerful AI Scientist released yet. We are a spinout from FutureHouse, focused on building and commercializing AI agents for science. Try our newest agents, including Kosmos, on our platform today.

[Daniel Burkhardt @DBBurkhardt]

@cohenrap @KrishnaswamyLab @scottgigante @Yale @fabian_theis @MDLuecken Unfortunately, Nature Biotechnology doesn't have the option for open access for commentaries, something we only realized post-review when we got to proofing the camera-ready version.

[Caleb Lareau @CalebLareau]

What a PERFFect day to share our new preprint!

With @tsion_abay #BobStickels @MerilTakizawa @ChaligneRonan and @Satpathology, we introduce PERFF-seq, a new experimental approach to studying rare cells with scRNA-seq via transcript-specific enrichment. https://t.co/woKvdEdOf0 1/n