Sources
- AI at Meta / @AIatMeta
- Amazon Web Services / @awscloud
- Anthropic / @AnthropicAI
- Cursor / @cursor_ai
- Google / @Google
- Google Cloud Tech / @GoogleCloudTech
- Google DeepMind / @GoogleDeepMind
- Grok / @grok
- Hugging Face / @huggingface
- Microsoft / @Microsoft
- OpenAI / @OpenAI
- OpenClaw🦞 / @openclaw
- Sequoia Capital / @sequoia
- Tesla / @Tesla
- Twitter / @a16z
- Waymo / @Waymo
- xAI / @xai
- Y Combinator / @ycombinator
Company@X — 2026-07-13#
Signal of the Day#
AWS has officially released the Agent Toolkit for AWS, allowing AI agents to securely connect to over 15,000 API actions via the Model Context Protocol (MCP). This is a massive step for automated cloud operations, granting AI agents scoped IAM access to handle deployments, configurations, and troubleshooting directly within enterprise environments.
Key Announcements#
Hugging Face · Source Hugging Face Transformers models can now run in vLLM at native speed without requiring a separate, hand-written implementation. This architectural unification eliminates the need to build models twice—once for training and once for inference—significantly accelerating the deployment of open-source AI into production.
OpenAI · Source ChatGPT has been reinstated on WhatsApp within the European Economic Area (EEA), alongside new availability on Kakao in South Korea and Viber. Users can natively interact with a verified ChatGPT contact to upload images, send voice notes, and generate content directly within these dominant regional messaging apps.
Waymo · Source Waymo signaled its readiness to launch commercial autonomous ride-hailing in Washington, D.C., backed by over 300,000 autonomous miles driven in the district. The company is preparing to invest tens of millions in local infrastructure and jobs, marking its next major metropolitan expansion.
Google Cloud · Source Google Cloud launched the Gemini Enterprise Agent Ready (GEAR) program, promoting production-grade multi-agent systems built with the Agent Development Kit (ADK), Gemini 3 Flash, and MCP. Developers can now design architectures and publish their A2A Agent Cards directly to the Google Cloud Marketplace to reach enterprise users.
Anthropic · Source Anthropic published new interpretability research demonstrating that Claude’s expressed values (such as warmth versus rigor) shift noticeably depending on the model version and the user’s language. For instance, Claude leans toward warmth in Hindi and Arabic, but toward rigor in Russian, highlighting unresolved challenges and behavioral nuances in cross-lingual AI steering.
Google DeepMind · Source DeepMind highlighted the “Predicting the Past” Skill within Google Antigravity, an AI-powered historical research workflow. The system uses a generative model named Aeneas to date, place, and analyze ancient texts in parallel, allowing researchers to reconstruct historical networks without needing to write code.
Also Noted#
- Little Tech Association (Source): A new trade group, backed by Y Combinator, Replit, and Proton, has launched to represent startups and advocate for merit-based tech competition over Big Tech market consolidation.
- Tesla (Source): Full Self-Driving (FSD) V14.3.5 introduces an always-available Camera Preview feature while driving, which includes a live feed of the interior cabin.
- Hugging Face (Source): The platform has opened ZeroGPU demo creation to all users via a new agent skill, democratizing access to GPU-backed application prototyping.
- InversionSpace (Source): The startup applied physical AI and computer vision to model parachute canopy deflections, successfully reducing autonomous landing errors by 90%.