Physical AI took center stage with Shanghai hosting the largest WAIC yet and Galaxy pushing embodied “soul” robots at the UN summit, while Veea launched a sovereign data platform for enterprise grids. OpenAI shipped GPT-5.6 into M365 Copilot and killed off GPT-5.2, Thinking Machines dropped its first open model, and the White House moved on voluntary frontier standards. Agent and identity startups kept raising big.
Top Stories
Shanghai to host biggest AI conference
SHANGHAI will host the largest World Artificial Intelligence Conference (WAIC) since 2018, pulling in more than 1,100 companies, researchers, policymakers and leaders. The scale signals China’s intent to dominate the global AI conversation. https://www.manilatimes.net/2026/07/16/news/national/shanghai-to-host-biggest-ai-conference/2385327
“The Future of AI is Soul” — Galaxy Corporation CEO at UN ‘AI for Good’ Summit
Galaxy CEO Yong-ho Choi closed the UN summit with a Physical AI keynote and K-POP humanoid robot performances, arguing embodied systems need “soul.” The event frames the next wave as hardware-tied rather than pure software. https://www.manilatimes.net/2026/07/16/tmt-newswire/pr-newswire/the-future-of-ai-is-soul-galaxy-corporation-ceo-yong-ho-choi-graces-the-finale-of-un-ai-for-good-summit/2385592
Veea Announces VeeaONE Distributed Intelligence Platform
Veea released VeeaONE middleware, Secured Docker toolkit and developer portal to build cybersecure sovereign data fabrics and enterprise AI grids aimed at Physical AI workloads. It positions local control and security as table stakes for scaled deployments. https://www.manilatimes.net/2026/07/16/tmt-newswire/globenewswire/veea-announces-the-veeaone-distributed-intelligence-platform-enabling-cybersecure-sovereign-data-fabrics-and-enterprise-ai-grids-for-physical-ai/2385564
OpenAI adds GPT-5.6 to Microsoft 365 Copilot
OpenAI made GPT-5.6 the default in M365 Copilot across Word, Excel, PowerPoint, Chat and Cowork for stronger enterprise drafting and analysis. At the same time GPT-5.2 variants were removed from ChatGPT in favor of the 5.5 line. https://releasebot.io/updates/openai https://help.openai.com/en/articles/6825453-chatgpt-release-notes
Thinking Machines releases open model Inkling
Thinking Machines shipped its first public open model after 18 months of quiet infrastructure work, explicitly rejecting one-size-fits-all approaches. The release tests whether specialized open models can carve out real usage. https://techcrunch.com/2026/07/15/thinking-machines-amps-up-its-bet-against-one-size-fits-all-ai-with-its-first-open-model-inkling/
White House advances voluntary frontier AI standards
The White House is in advanced talks with OpenAI, Google and Anthropic on voluntary release standards, with an announcement expected soon. The move aims to shape frontier behavior without hard regulation. https://aitoolsrecap.com/Blog/AINewsJuly2026.aspx
Emergent joins unicorn club after $130M round
Emergent raised $130M in Series C to become India’s latest AI unicorn. Parallel rounds included Oak’s $60M seed for an AI-native identity OS and Trase’s $107M for healthcare AI agents. https://apacnewsnetwork.com/2026/07/emergent-joins-unicorn-club-after-130-million-funding-round/
Captain’s Take
Physical AI and sovereign grids are the real through-line. VeeaONE and the Shanghai conference both point to enterprises wanting locked-down, location-aware fabrics. That aligns with Mate’s local-first stance. The “soul” robot demos are theater until someone ships private orchestration that actually runs on customer hardware without phoning home.
OpenAI’s weekly model shuffle keeps proving the point: never bet the product on a single checkpoint. We abstract and swap. The voluntary standards chatter is background noise until it creates real deployment friction or moats. Watch what actually ships out of WAIC and whether any of the new open models like Inkling survive contact with real workloads.
Funding keeps validating the implementation layer. Agent and identity rounds show the market wants the messy work of making AI fit actual lives and companies. Our edge is still the door-to-door setup and ongoing local tuning. Keep shipping the personal agents that run on Captain’s stack while everyone else chases the next conference headline.