Coinbase agents can now utilize the x402 protocol to gain direct access to exclusive data and APIs, enhancing their ability to trade and pay for premium research insights.

Alt Carbon revealed that the partnership with Microsoft came after over a year of rigorous scientific evaluation and thorough due diligence. Microsoft insisted on extra verification and comprehensive data-sharing protocols to ensure transparency and accuracy.

THEKER, an AI robotics firm based in Barcelona, has successfully raised €73 million ($85 million) in a Series A funding round aimed at expanding its versatile factory robots in industrial production settings. The funding round was led by CRV and included key participants such as Samsung, LVMH, Cathay Innovation, 20VC, Henkel Ventures, Korelya, and Bright Pixel Capital. Notably, this marks Samsung’s inaugural investment in a Spanish tech company.

Prometheus, the AI startup co-founded by Jeff Bezos, has successfully raised $12 billion in its latest funding round, bringing its total valuation to $41 billion. Major investors in this round include JPMorgan Chase, Goldman Sachs, BlackRock, DST Global, and Arch Venture Partners, alongside Bezos. With total funding now topping $18 billion, Prometheus aims to develop advanced artificial general intelligence designed to engineer physical products, marking a major leap in AI-driven innovation.

Which of the emerging obesity drugs has the potential to truly lead the market? Which company recently made headlines with a record-breaking IPO? And what decisions might the Food and Drug Administration be making about these treatments?

Innovation has become a common buzzword, often losing its real meaning through overuse. When asked to define it, many executives offer vague and unhelpful answers. The truth is innovation is challenging and celebrated more in hindsight than practiced in the moment. Three key secrets stand out:

1. Innovators have a distinct mindset or phenotype, defined by a willingness to question norms and accept calculated risks. This mindset can't simply be installed—it flourishes when the company culture supports risk-taking and learning from failure.

2. Speed outweighs certainty in decision-making. Rapid decisions with incomplete information often trump slow, perfect choices. In early stages, focusing less on competitors and more on building what should be is crucial.

3. Failure isn’t the enemy—it’s necessary for progress. The key lies in failing wisely: quick, small, and insightful experiments that lead to learning without draining resources. Companies that manage failure well foster a culture of safety where insights, not defeat, come from setbacks.

Ultimately, innovative companies are agile, embrace risk strategically, and hire individuals with the right innovative traits to push boundaries continuously.

Agent skills, which are sets of instructions stored as markdown files, allow AI models to adapt to specific tasks and enterprise workflows. Traditionally, optimizing these skills is a slow, manual process plagued by trial-and-error. Microsoft’s new open-source framework SkillOpt revolutionizes this by treating skill documents as trainable objects that evolve based on performance feedback, using deep-learning-inspired optimization techniques—without changing the underlying AI model's weights. It iteratively proposes and tests edits to skill documents, ensuring improvements are validated mathematically before being accepted. Testing across various models, including GPT-5.5 and Qwen, showed significant accuracy improvements and adaptability across different execution environments. SkillOpt produces compact, reusable skill artifacts that improve AI reliability in complex workflows like document data extraction and tool use automation. This framework is cost-effective, efficient, compatible with existing systems, and emphasizes procedural learning over memorization, making it highly valuable for enterprise applications.

Long context windows in large language models (LLMs) increasingly strain computational resources, as accumulated tokens from documents, conversation histories, and reasoning processes grow. Traditional compression methods either reduce model accuracy or fail to speed up real-world applications effectively. A research collaboration from NYU, Columbia, Princeton, University of Maryland, Harvard, and Lawrence Livermore National Laboratory introduces Latent Context Language Models (LCLMs), a breakthrough encoder-decoder design that compresses the input context before decoding, significantly cutting memory and compute demands. LCLMs achieve up to 16x compression, making inference nearly 9 times faster on long-context benchmarks while maintaining competitive accuracy compared to existing methods. The architecture combines a lightweight encoder with a larger decoder, trained on a vast dataset with a novel mix of continual pre-training, fine-tuning, and reconstruction tasks to balance compression and performance. Integrated easily into existing pipelines, LCLMs allow models to handle much longer contexts more efficiently and cost-effectively. This advance offers enterprises a powerful tool to manage growing context lengths without sacrificing speed or accuracy, though careful tuning for retrieval-augmented generation systems remains necessary. The models and code are openly shared on HuggingFace and GitHub.

Generative AI image models like Stable Diffusion refine images by iteratively denoising noise across the entire canvas, a technique called diffusion. Applying this approach to text generation at scale had long been challenging. Traditional language models generate text sequentially, one token at a time, limiting speed and flexibility. Google's new open-source DiffusionGemma model changes this by generating 256 tokens simultaneously, refining them through multiple passes to self-correct mistakes. Built on the Gemma 4 architecture and integrated with the vLLM inference platform, DiffusionGemma can produce text up to 4 times faster on GPUs, reaching over 1,000 tokens per second on Nvidia H100 and over 1,200 on H200.

Unlike typical models, DiffusionGemma generates tokens in parallel and revises uncertain outputs with each pass, enabling bidirectional context that benefits constrained text tasks like coding or puzzle solving. It achieves this efficiency by activating a fraction of its 26 billion parameters per inference, fitting consumer GPUs like the RTX 4090. However, this speed gain comes with a trade-off in output quality compared to standard Gemma 4, making it less suitable for applications demanding the highest text fidelity.

This approach excels in scenarios with single-user or low-concurrency inference where GPUs are underutilized but is less advantageous in high-throughput cloud environments where sequential models saturate GPU resources. DiffusionGemma introduces a structural shift in text generation, offering a promising new option for enterprises needing fast, bidirectional, and flexible text generation on local or dedicated hardware setups.

The narrative around artificial intelligence is polarized and overwhelming. On one side, AI is portrayed as a catastrophic threat; on the other, it's hailed as a revolutionary force for progress. Yet, neither the dystopian fears nor the utopian promises fully capture reality. AI's impact is massive, already driving significant economic growth—by late 2025, it contributed to nearly 60% of the U.S. economy's expansion. Despite its potential, the future of AI is not predetermined, and the hype often obscures a more nuanced understanding of its capabilities and risks.