In partnership with
Welcome back to The AI Field. Google is turning game ideas into something you can play, Biohub is investing in AI models of living cells, and Grok Bot is opening up to rival models.
Today: what you can try now, what is still being built, and four useful tools to explore.
In today’s AI Field:
Google’s Playground turns your game idea into something you can play
Biohub’s $1.8B push to build AI models of human cells
Grok Bot will use Claude, Midjourney and Suno
Plus: free security scans for open-source projects, Nous Research’s $90 million raise, and Arena’s new test of whether AI agents follow your intent.
Read time: 5 minutes
Master Claude AI (Free Guide)
The professionals pulling ahead aren't working more. They're using Claude.
Our free guide will show you how to:
Configure Claude to be the perfect assistant
Master AI-powered content creation
Transform complex data into actionable strategies
Harness Claude’s full potential
Transform your workflow with AI and stay ahead of the curve with this comprehensive guide to using Claude at work.
LATEST DEVELOPMENTS
GOOGLE LABS · GAME CREATION
Google Labs has launched Playground, an experimental game-making tool that turns a description into a playable game. You build through a chat interface, so you can start with an idea instead of a game engine.
The details:
Start from scratch or choose a genre such as trivia, tower defense or racing. Describe the game, play it, then ask for changes.
You can browse community creations and share your own games with friends or the wider Playground community.
The launch is for US users aged 18 and over. Playing is free. Free creation access is rolling out through a waitlist with weekly tokens, while Google One AI subscribers get immediate creation access and higher limits.
The AI Field: A playable prototype tells you more than a page of game ideas. This is worth trying for a small quiz, a teaching exercise or a quick experiment with friends. Judge the game by whether it is fun to play, not just how impressive the first prompt looks.
BIOHUB · AI FOR SCIENCE
Image: Biohub.
Biohub is expanding its Virtual Biology Initiative with US government agencies and major AI companies. The goal is to collect the data needed for AI models that predict how cells respond to drugs and other changes.
The details:
Google DeepMind, Isomorphic Labs and Meta are committing $300 million together.
The Department of Energy will contribute more than $500 million over five years. NIH will supply datasets developed through more than $500 million in earlier federal spending.
The effort builds on Biohub’s original $500 million commitment. The $1.8 billion total combines funding, existing data, computing and technology. It is not all new cash.
The AI Field: Better cell models would let scientists test more ideas digitally before running lab experiments. The hard part is collecting enough useful biological data. This is an investment in that foundation, not the arrival of a finished virtual human cell.
TLDR AI is a free daily email curated by Anthropic and ex-Google engineers: the new tools worth trying and the research worth understanding, summarized for 1.1M+ readers.
GROK BOT · AI AGENTS
Elon Musk says Grok Bot will choose outside AI models when they are better suited to a task. That includes tools made by rival companies.
The details:
Musk explicitly named Claude Opus 5.5, Midjourney and Suno, alongside other leading APIs.
His stated priority is picking the backend most likely to produce the best result for the task.
The post does not give a rollout timetable, pricing or details about how tasks will be assigned. It concerns Grok Bot, not a claim that the entire Grok chatbot is being replaced.
The AI Field: For an agent, choosing the right model is part of doing the job. A useful assistant should know when to hand work to a specialist. The test is whether that makes the result better without making cost and data access harder to understand.
TRENDING AI TOOLS
📊 Claude Dashboards: Turn connected company data into live dashboards and inspect the query behind each number. Useful when you want to check where a chart’s answer came from. Beta on paid plans.
🎬 HyperFrames Studio: Build videos with an AI agent, then refine them together on a shared timeline. For creators who want to keep editing the result instead of stopping at a generated clip. Available for Mac and Linux.
📞 ElevenAgents Architect: Describe the support agent you need, then diagnose weak conversations and draft improvements. Proposed changes require your approval before going live. Available in alpha.
📣 Runway Ads: Generate image and video ad variants, approve what gets published, and use campaign results to guide the next batch. A new workflow for teams testing creative across campaigns.
QUICK HITS
🔐 Anthropic offers free security scans for open source: Its new OSS Scanner gives eligible projects recurring vulnerability checks. Reports arrive without human review, so maintainers still need to verify the findings before acting on them.
💸 Nous Research raises $90 million: The company is expanding Hermes into business use, with customers able to choose the models behind their agents. The pitch is more control over how an agent is built and run.
📊 Arena raises $200 million and adds an Alignment Index: The AI evaluation platform is now valued at $3.1 billion. Its new index is designed to test how reliably agents follow human intent, adding another way to judge them beyond which answer wins a comparison.
That's it for today!
See you soon,
Olle Hellman | Founder of The AI Field