Triatomic Capital

Intelligence will soon be commoditized. The thing that will not be commoditized is data. Nowhere is data more important than in science. Karp's words are important for all companies, but especially for companies in science: you have to make sure you own your data.

Name your metrics. Get a filterable dashboard that updates itself. Connected to your live data. Filterable by any dimension. No tedious setup required. Most dashboards are built once and go stale. Someone exports a CSV, pastes it into a slide, and by the time the meeting starts the numbers are wrong. Everyone knows it. Nobody fixes it. This one doesn't work that way. You tell Coworker what you want to track. It builds the dashboard, wires it to your live data, and keeps it current. Filters for region, segment, time period. Metrics that refresh without anyone touching them. 1,243 active accounts. 38 at-risk. Avg health score of 74. All visible in one place, updated automatically. The dashboard doesn't replace the analyst. It frees them from the part of the job that was never analysis. What metric does your team check manually every week? coworker.ai

The future of AI inference is rack scale. A full rack of SMC X-14 servers powered by d-Matrix Corsair™ accelerators and connected with JetStream™ networking. Purpose-built for high-performance, low-latency, power-efficient AI inference. Real infrastructure. Real engineering. Real systems. And yes...real racks have wires. 😉 #AI #AIInference #RackScale #AIInfrastructure #HPC #DataCenter #dMatrix

Exploring chemical space with Assembly Theory.

We're excited to announce that @parasail_io is deploying d-Matrix Corsair™ inference accelerators alongside @nvidia Hopper and Blackwell GPUs to power heterogeneous AI inference. By pairing NVIDIA GPUs for compute-intensive prefill with Corsair for latency-sensitive decode, Parasail can deliver up to 10x faster interactive token generation with improved cost-performance and energy efficiency. The future of AI inference is heterogeneous. Read more: d-matrix.ai/announcements/… #AI #Inference #LLM #GenAI #NVIDIA #AIInfrastructure #dMatrix @Parasail_io

Introducing Open Artifacts by Coworker. Build beautiful work products, routed to the right model for every task. Frontier outputs. Exactly the right context. 80% lower cost.

New from our team at Gimlet: formal verification for AI-generated (and hand-written!) kernels. We lower the PyTorch reference and generated kernel to a common representation and ask Z3 to find an input where they disagree - catching structural bugs missed by typical numeric tests. gimletlabs.ai/blog/formally-…

Advances in AI coding / ability to produce software has exposed the chasm between the hardware world and the software world. It is time to invest into making matter programable. My company @ChemifyX & my research team @UofGlasgow using chemputation is leading the charge.

New Electronics covered the launch of Efficient Labs, our free suite of browser based tools for the Electron E1 evaluation kit. From board visualization to energy profiling, the tools are designed to cut through fragmented docs and manual configuration. Read the full piece here: newelectronics.co.uk/content/news/e…

Our paper on Robin came out in Nature yesterday -- awesome to see it in print finally. Congrats to the team! @agreeb66 @MichaelaThinks @andrewwhite01 @benjamin0chang and many others.

Edison Scientific and investment firm Population Health Partners are teaming up to leverage AI agents in drug discovery and development. trib.al/KfHJNDa

At Edison, we are on a mission to leverage AI to prevent or cure all diseases. Today, we are announcing a critical step: our first partnership using our AI scientist, Kosmos, to launch new biotech companies. Many of the most transformative new medicines begin life in biotech companies rather than big pharmas. We are partnering with Population Health Partners (PHP) to deploy Kosmos across their company creation pipeline. Kosmos will assist in the creation of PHP’s companies, allowing us to bring more medicines to patients. PHP has an outstanding track record of incubating successful biotech companies like @MetseraInc (acquired by @pfizer in a $10B deal), and The Medicines Company (acquired by @Novartis for $9.7B), to name just two. With PHP, we will focus on biotechs treating population-scale diseases like cardiovascular disease and obesity. If we can scale up the process of launching Metsera-style companies, we have an opportunity to transform medicine in the coming years.

If you want to work on using AI to accelerate the discovery of new medicines and scale up the creation of new biotech companies, get in touch.

Announcing the first production robot navigation framework on $500 hardware Explore the world once → your robot agent will relocalize and build a persistant, spatial memory across sessions SLAM, relocalization, loop closure, map i/o, planning, control No ROS. Open source.

Open source robotics with navigation / persistent spatial memory. Get it: @dimensionalos

The effcc Compiler drops into your existing workflow. No rewrites. No new toolchains. The effcc Compiler maps standard C and C++ onto the Fabric architecture at compile time, executing operations concurrently across tiles as the application runs. Download the product brief: efficient.computer/effcc-brief #ElectronE1 #EfficientCPU #EdgeAI #DeveloperTools

At Cornell Tech's inaugural Frontiers of AI Summit, Gimlet cofounder @nserrino spoke about the current state of inference and why agentic inference requires heterogeneous hardware. Link in reply!

We're humbled to share that Corsair just won the AI Processor Innovation Award as part of this year's AI Breakthrough Awards. The industry is moving swiftly to heterogeneous compute. We built d-Matrix exactly for this shift and have begun shipping in volume. 🔗 d-matrix.ai/announcements/… #AI #Inference

The fastest way to understand a new architecture is to see your own code running on it. The Efficient Computer Playground is a browser-based visualization tool. Paste your C code, compile it with the effcc Compiler, and watch how it maps across the Fabric architecture, tile by tile, cycle by cycle. No hardware required. Try it: playground.efficient.computer/login #ElectronE1 #EfficientCPU #EdgeAI

.@RadicalNumerics came out of stealth today with the mission to be the leading AI research lab with a dual mandate across human health + bio-defense. Super excited for @TriatomicCap to be working with @exnx, @MichaelPoli6, @athmsx, @Massastrello & Team (builders of Evo/Evo2). More from Fortune: fortune.com/2026/06/15/exc… More: radicalnumerics.ai/blog/radical-n…