Welcome to the Age of Machine Scientists
For the first time, research agents that rival top scientists are available to everyone.
Good morning AI entrepreneurs & enthusiasts,
Scientific research is about to hit a new gear, with FutureHouse unveiling four advanced AI agents that reportedly outperform PhD-level researchers.
With agent-based research assistants now able to comb through millions of studies and datasets, a surge of AI-powered scientific discovery is closer than ever.
In today’s AI news:
FutureHouse’s breakthrough science agents
Apple & Anthropic team up on AI coding platform
Google addresses AI's energy and labor crunch
David Sacks' 1,000,000x AI projection
Top Tools & Quick News
FutureHouse introduces next-gen AI science agents
The News: Backed by Eric Schmidt, FutureHouse just launched a suite of AI-powered research agents built for accelerating scientific discovery. These tools aim to eliminate the data overload researchers face by simplifying complex information retrieval.
The details:
Four agents are available now: Crow (general research), Falcon (deep literature review), Owl (research traceability), and Phoenix (chemistry-specific workflows)
The agents outperform PhD researchers and current SOTA models in literature analysis and synthesis
They can access niche databases and show transparent reasoning paths for every insight
Available via web interface and API
Why it matters: While many AI science platforms are still in development, FutureHouse’s tools are already live and in use, representing a foundational shift in the role AI plays in human advancement. This isn’t just about efficiency — it’s about rewriting the limits of what's scientifically possible. By embedding autonomous reasoning into research workflows, we’re moving toward a future where knowledge discovery is not just accelerated but reimagined entirely. The age of the AI-augmented scientist has arrived — and with it, the beginning of an era where machines and humans co-create the future of innovation.
Apple and Anthropic join forces on AI coding tools
The News: Apple is partnering with Anthropic to build a new "vibe-coding" platform for Xcode, its integrated development environment. The tool will auto-generate, edit, and test code using Anthropic’s Claude Sonnet model, allowing developers to interact with AI in natural language.
The details:
Internal testing is underway; a public release date is yet to be announced
Developers can chat with Claude Sonnet to write, edit, debug, and test code
Apple plans to integrate Google Gemini later this year to broaden its AI stack
Why it matters: Apple is signaling a strategic shift from secrecy to pragmatic collaboration by integrating external AI technologies into core platforms like Xcode. But despite the partnership’s optics, many in the developer community remain underwhelmed. With no public release date, limited functionality, and a feature set that merely meets the baseline for what an IDE-integrated LLM should do, Apple appears to be playing catch-up in a space that demands boldness. While competitors like Microsoft have already operationalized AI coding tools at scale, Apple’s slow and cautious rollout feels more like a defensive maneuver than a visionary leap forward.
Google’s new energy & workforce strategy
The News: Google released a comprehensive policy roadmap and workforce plan to address AI’s growing power and labor demands.
The details:
The roadmap includes 15 proposals to modernize U.S. energy infrastructure through advanced energy tech, grid optimization, and improved transmission planning
A new partnership with the Electrical Training Alliance (etA), IBEW, and NECA will upskill 100K workers and add 30K new apprentices by 2030
This initiative expands Google’s AI Opportunity Fund to support not just digital skills, but now infrastructure and trades essential to data center growth
Why it matters: The AI boom is creating real strain on energy and labor markets. Google’s move sets a new precedent for Big Tech’s role in not just consuming infrastructure, but building it. It reflects a long-term, systems-level approach to ensuring that AI growth doesn’t outpace the resources and workforce needed to support it.
David Sacks projects 1,000,000x AI scale-up in 4 years
The News: U.S. AI czar David Sacks predicts AI capabilities will grow by 1,000,000x over the next four years, driven by exponential improvements in algorithms, chips, and compute infrastructure.
The details:
Sacks notes that each new model generation is 3–4x better than the last—moving rapidly from chatbots to reasoning models and soon to autonomous agents
Nvidia’s NVL72 racks and innovations like 3D stacking offer up to 30x improvements in throughput over H100s; the upcoming GB300 NVL72 could provide a 10x boost in responsiveness
OpenAI, xAI, and others are scaling from thousands to millions of GPUs, enabling larger models to be trained and deployed
The combined effect, Sacks explains, is a 10x improvement in each domain every 2 years—which compounds to a 1,000,000x leap in capability
Why it matters: If this projection holds, we’re on the brink of a transformation beyond anything society has experienced. Superhuman AI could revolutionize science, medicine, and engineering—but also introduce unprecedented challenges in power demand, reliability, and governance. As Sacks warns, exponential progress doesn’t double—it compounds.
Today's Top Tools
Phi-4: Microsoft’s new lightweight reasoning model
Nova Premier: Amazon’s teacher-tuned multimodal model
Midjourney Omni Reference: Add consistent characters via image prompts
Quick News
Gemini 2.5 Pro just beat Pokémon Blue live on stream — Claude still can’t
Anthropic may buy back employee shares at a $61.5B valuation
Gemini for kids under 13 is rolling out under Family Link protections
DeepMind teases 10M-token windows for superhuman codebases
Zoom’s new "Chain of Draft" method matches CoT accuracy with 7% tokens
Thanks for reading this far! Stay ahead of the curve with my daily AI newsletter—bringing you the latest in AI news, innovation, and leadership every single day, 365 days a year. See you in the next edition!