Zuckerberg swung into $500 million AI-based biology

Good morning {{first_name| Artificial intelligence lovers}}. Google Deepmind’s Demis Hassabis predicted that AI could eventually end disease, and Mark Zuckerberg and Priscilla Chan Biohub just placed $500 million on the same bet.

The new five-year virtual biology initiative combines significant funding with a strong group of partners, all coordinated around the same goal: producing enough data to drive artificial intelligence to model how disease begins at the cell level.

reminder: Our next live workshop is today at 2pm EST! Join our walkthrough of OpenAI’s Codex platform, its new features, and how to get the best out of it as a non-technical user. RSVP here.

In today’s AI summary:

• Zuckerberg’s Biohub is funneling $500 million into AI biology

• Mayo Clinic AI detects pancreatic cancer 3 years earlier

• Build a custom blog writing agent without code

• Food AI’s ChatGPT moment – Tastes like a chef

• 4 new tools for AI, community workflows, and more

Latest developments

BioHub

🧬 Zuckerberg Biohub andInvests $500 million in AI biology

Image source: Images 2.0 / Rundown

Rundown: Biohub, a nonprofit backed by Mark Zuckerberg and Priscilla Chan’s CZI, announced a $500 million Virtual Biology Initiative to build open datasets and models that can predict how human cells will behave – pushing AI toward mimicking biology.

the details:

• Of the $500 million, $400 million will fund data generation and imaging technology, with $100 million for outside research laboratories and research efforts.

• Nvidia, Allen Institute, Arc and others are joining the initiative, with Biohub committed to opening up datasets as a common base for AI biology research.

• Current AI biology datasets max out at near 1B cells, with Biohub’s Alex Rives saying more data is needed to accelerate efforts.

• The goal is to train models on data to use AI to “understand and reprogram disease at the cellular, molecular and tissue level.”

Why it matters: Google’s Demis Hassabis has said that AI could eventually end disease, and Biohub is pouring big money into the same line of thinking. The question is whether the analogy that cracked language and protein structure also applies to cells, and whether the $500 million cost comes close to the amount of data needed to find out.

Along with Glenn

📈 From agent extension to real ROI

Rundown: Organizations are deploying agents across functions, but few are seeing the ROI they expected. Join Glean:LIVE on May 12 to discover the enterprise agent development lifecycle – a new operating model that moves organizations from dispersed experiences to agents that deliver real, measurable impact.

Register on Glean: Direct to:

• Get a repeatable framework for building, launching, and managing agents at scale

• Hear from organization leaders about the decisions shaping their dealership strategy

• Watch product demos covering the full agent lifecycle – with a live QandA at the end

Register now.

Artificial intelligence and healthcare

🔬 Mayo Clinic AI detects pancreatic cancer 3 years earlier

Image source: Mayo Clinic

Rundown: Mayo Clinic has published new data on REDMOD, an artificial intelligence system that reads tissue patterns not visible in standard CT scans, detecting pancreatic cancer up to three years earlier than doctors would normally detect it, nearly doubling specialized accuracy.

the details:

• REDMOD reviewed nearly 2,000 routine CT scans, which specialists originally read as normal before subsequent diagnosis, and caught 73% of cases early.

• Two years after diagnosis, the gap had widened even more, with the AI ​​detecting nearly three times as many early cancers as experienced radiologists.

• The form reads andquot;Hundreds of quantitative imaging features,andquot; Texture and structure patterns are usually invisible to human radiologists.

Why it matters: The 5-year survival rate for pancreatic cancer is less than 15%, making early diagnosis and treatment critical. With REDMOD running on scans that patients already get, AI’s early screening capabilities could become a standard part of routine care rather than a separate diagnostic step that adds friction to the system.

Artificial intelligence training

✍️ Build a custom blog writing agent without code

Rundown: In this guide, you’ll learn how to create a sub-agent in Langflow that writes blog posts in the style of your website. The big win is that the proxy runs locally for free (besides LLM calls) and can be run by Claude, Codex, etc.

Step by step:

1. Download Langflow, open it on your computer (runs locally), click New Flow, go to the Content Creation section, and choose the Blog Writer template

2. Add a text entry called andquot;addressandquot;giving the agent a topic to write about, and linking your best blog in the reference entry so the agent can copy your writing style

3. Add an OpenAI or Anthropic API key under Language Model Settings. Then select the form you want the flow to use

4. Click on Playground and test it with a theme. This turns on the flow, captures your reference style, and gives you a first draft

Pro Tip: If you want Claude to use this as a subproxy, click Share > MCP Server > Claude MCP. Then Claude can use this blogging tool whenever you need it.

Provided by Box

🧠 Transform institutional knowledge into AI-driven action

Rundown: Enterprise AI only works when it has the right business content. Box is a secure, essential context layer for agents to access the institutional knowledge that helps the company run.

Box’s AI suite includes:

• Box Extract to pull actionable data from enterprise content at scale

• Box Agent to transform unstructured data into the context that AI needs

• Box Automate (now GA) for organizing agent workflow automation

Learn more about Box’s AI solutions At Box’s Content + AI Virtual Summit on May 20.

Artificial intelligence research

🧑‍🍳 Food AI’s ChatGPT moment – Tastes like a chef

Image source: mid-flight

Rundown: Food robotics startup KAIKAKU AI Just published Epicur, a new paper claiming A andquot;ChatGPT momentandquot; For food, AI shows that its AI model can capture flavour, cuisine and texture just from how chefs combine ingredients in recipes.

the details:

• The researchers filtered through 6,653 messy ingredient entries in 1,032 usable foods, then mapped how they related across recipes using artificial intelligence.

• Although you never saw chemistry data or taste labels, the model identified all five basic tastes, ranked peppers by spiciness, and labeled cuisines by region.

• The team identified three applications: menu development, recipe innovation, and flavor pairing, work typically driven by the chef’s intuition.

• KAIKAKU pairs the AI ​​from this paper with its robotic arm, rendering the narrative as is andquot;Independent food infrastructureandquot; For commercial kitchens.

Why it matters: Recipes are already powerful data points for human preferences, with every pairing, permutation, and pattern indicating what people think works. If AI can read this structure, it can also help implement tools to design menus, suggest better alternatives, and create products with actual taste and texture in mind.

Fast strokes

🛠️ Popular AI tools

• 🎶 ElevenMusic – A platform for AI song generation, remixes, and creator compensation

• 🚀 MiMo-V2.5-Pro ​​- Xiaomi’s new powerful open source model

• 🧠 GPT 5.5 – OpenAI frontier model for proxy encryption and computer use

• ⚙️ Cursor SDK – Run Cursor coding agents in workflows and other products

📰 Everything else in artificial intelligence today

Eleven laboratories It launched ElevenMusic, a streaming platform with built-in AI remixes and AI-assisted track creation, that already hosts 4k+ artists and offers payments to creators.

Two committees in the US House of Representatives Investigations opened into index maker Anysphere and Airbnb over use of Chinese AI, with Composer 2 built on Kimi and Airbnb’s proxy Qwen.

Mistral AI It launched Vibe Remote Agents, cloud sessions that run programming tasks in parallel, powered by the company’s new open-weight model Medium 3.5.

Google File generation has been added to Gemini, allowing the model to output formats such as Google Docs, Sheets, Slides, Microsoft Word and Excel files, Markdown, and more.

OpenAI Issued a cybersecurity action plan andquot;Democratizationandquot; AI cyber defense and working with the US government and industry to coordinate threats and defense tools.

community

🤝 Community AI workflow

In each newsletter we show how the reader is using AI to work smarter, save time or make life easier.

Today’s workflow comes from the reader Patrick B. In Boonsboro, Maryland:

andquot;A few months ago, I received a cancer diagnosis…the number of appointments, trying to understand the medical concepts, the diagnosis, the appointments, the tests, how cancer really works, is truly overwhelming.
I’ve built several assets and artifacts in Claude, including a dashboard to help me manage my diagnosis through appointments, insurance claims, screenings, treatment, regimens, etc. This also allowed me to share with my family the progress and what was happening with my diagnosis. As much as I am a patient advocate, this has made it much easier for me to ask better, more targeted questions to my oncology team and has helped me manage this entire process much better than I could have done otherwise.
This has been a huge advantage for me in tracking my diagnosis and making sure I follow my doctor’s instructions in a more detailed manner. It also gives me a central place to ask questions about my diagnosis because he knows me as well as my doctors.andquot;

How do you use artificial intelligence? Tell us here.

🎓 Highlights: News, guides and events

• Read our latest newsletter on AI: The largest AI experiment ever launched

• Read our latest tech newsletter: Samsung’s Meta Ray-Ban competition just leaked

• Read our latest newsletter on bots: China’s bot ban is coming

• Today’s AI Tools Guide: Build a Custom Blog Writing Agent Without Code

• RSVP to today’s workshop at 2pm EST: Codex for Non-Technical Operators

see you soon,

Rowan, Joey, Zack, Shubham and Jennifer – the humans behind The Rundown