🤖 OpenAI and Anthropic bet $11.5B on enterprise AI 🤖
Two rival AI labs launch parallel PE-backed ventures for enterprise rollout.

Anthropic announced a $1.5 billion joint venture with Blackstone, Goldman Sachs, Hellman & Friedman, and Sequoia to deploy Claude across mid-sized companies. Hours later, OpenAI unveiled "The Deployment Company," a $10 billion joint venture with Bain Capital, Brookfield, TPG, and SoftBank to roll out its tools across their portfolio companies. Both initiatives give frontier AI labs direct commercial paths into enterprises that typically lack in-house talent to deploy AI — shifting competition from model performance to enterprise distribution channels. The parallel announcements mark a major strategic pivot: private equity firms are now the primary go-to-market engine for frontier AI, and the Anthropic-OpenAI rivalry has officially expanded well beyond the lab.

🚀 Cofounder 2 ships full AI teams for solopreneurs 🚀
A new platform replaces entire startup teams with orchestrated autonomous agents.

General Intelligence has released Cofounder 2, a platform that lets solo founders spin up and run entire companies staffed by autonomous agents across engineering, sales, marketing, and operations. The system provides a visual org chart of every agent and a company roadmap for setting goals and monitoring progress, effectively offering a full-stack AI team on demand. This pushes agent orchestration from single-task tools to full company-level automation, a meaningful signal for how small business operations may look in the near future. Cofounder 2 drew attention across multiple outlets for its ambitious all-in-one approach to agentic business integration.

🔬 Meta releases Tuna-2 with pixel embeddings 🔬
Meta's new multimodal model takes a different architectural path from token-based rivals.

Meta has released Tuna-2, a multimodal model that outperforms both Tuna-R and the original Tuna across a diverse suite of benchmarks by leveraging pixel embeddings instead of the token-based approaches favored by competitors. Meta is publishing a foundation checkpoint rather than full production-trained weights — a small number of layers have been removed from the LLM backbone and diffusion head, though all remaining components are fully preserved. The pixel-embedding approach offers a new architectural direction for joint vision-language models, and the repository on GitHub includes sample generated images. This signals Meta's continued commitment to open multimodal research, even with partial access restrictions on weights.

INTERESTING TO KNOW

⚖️ White House eyes pre-release vetting for AI models ⚖️

The Trump administration is actively discussing a potential executive order to create an AI working group of tech executives and government officials that would establish regulation procedures for reviewing AI models before public rollout. If enacted, pre-release oversight could reshape how labs like OpenAI, Anthropic, and Google launch frontier systems. The proposal follows the rescinding of Biden-era AI executive orders and comes amid rising concern about the speed of capability advances in autonomous AI systems.

⚙️ Google adds Webhooks to Gemini API ⚙️

Google has shipped event-driven Webhook support for the Gemini API, enabling developers to receive push-based notifications when long-running jobs — such as batch generation or fine-tuning — complete, replacing inefficient polling. The speed and integration improvement closes a key gap for production-grade agentic applications, bringing the Gemini API closer to feature parity with competing platforms.