OpenAI and Broadcom unveiled Jalapeño, a custom AI chip specifically designed to improve performance and efficiency in LLM inference. This development matters for developers looking to optimize costs and latency in production deployments.

Anthropic launched Claude Tag, enabling teams to collaborate with Claude directly in Slack. 65% of Anthropic's product team's code is already created using this feature, demonstrating its potential as a team workflow tool.

NVIDIA and AWS announced a collaboration to deploy AI systems in production at scale, addressing latency, vector search, and cost efficiency challenges. The initiative provides practical paths for enterprises to adopt AI in real-world environments.

Qualcomm completed its acquisition of Modular, a promising AI chip software startup, for nearly $4 billion. This deal reflects the accelerating race to control AI infrastructure and hardware optimization.

Claude Tag in Slack isn't just about productivity—it's a strategic play to capture organizational context, institutional knowledge, and enterprise workflows. Anthropic aims to become essential infrastructure for business operations.

MoEngage acquired technology that assigns individual AI agents to each customer for personalized marketing. The move reflects evolution toward agentic systems as the basis for enterprise strategies.

Stripe, Anthropic, and OpenAI are co-funding a research initiative to prevent respiratory infections. It exemplifies AI application beyond traditional software toward real biomedical challenges.

Researchers introduced RIFT-Bench, a methodology for evaluating agentic AI system security through dynamic red-teaming. Critical tool for developers deploying autonomous agents in production.

DeepSWE is a new benchmark evaluating how well frontier models write code with real complex tasks. The benchmark is free of contamination and more challenging than previous evaluations.

New EXPO-SQL method improves natural language to SQL translation using clause-level rewards instead of query-level rewards. Significant advance in RL applied to SQL code generation.

OpenAI participated in initiatives to build shared standards for evaluation, safety, and global cooperation in advanced AI systems. This effort aims to establish industry-wide practices to ensure reliability.

IBM Research and Hugging Face introduced CUGA, a lightweight framework with twenty working examples of agentic applications. The tool lowers the barrier for developers building autonomous AI systems.

NVIDIA published guidance on building specialized AI agents that fit real enterprise workflows. The proposal marks a transition from exploratory AI toward task-oriented autonomous systems.

Enterprises struggle to access web-scale data for AI systems, as much information is blocked or unstructured. A new infrastructure layer is emerging to solve this bottleneck.

Comparative pricing analysis across 7 LLM providers reveals prompt caching is more important than headline pricing. For agents, RAG, and reusable prompts, caching policies vary up to 10x in cost.