• AI Giants Back Nonprofit Focused on Workforce Transition — OpenAI, Anthropic, Microsoft, and Amazon are backing Raise US, a new nonprofit aiming to raise $1 billion to help American workers prepare for an AI-driven economy, signaling a shift from pure model-building toward workforce development. 🔗 Graph: OpenAI, Anthropic, Microsoft, AI Governance, Higher Ed AI 📅 Published: 2026-07-06 📰 https://campustechnology.com/articles/2026/07/06/ai-giants-back-nonprofit-focused-on-workforce-transition.aspx 📌 Key takeaways: • The nonprofit was founded by former U.S. Commerce Secretary Gina Raimondo and former Indiana Gov. Eric Holcomb; Bank of America has joined as primary corporate sponsor. • Raise US has already secured more than $500 million in commitments and will launch workforce initiatives in Arkansas, Connecticut, Maryland, and Utah. • "America has a technology strategy for leading the global AI competition. It does not yet have a people strategy — and we cannot lead without one," Raimondo said. • The initiative reflects a broader recognition that AI will dramatically reshape jobs, and that the private sector — not just government — needs to invest in reskilling and career development.

• Expanding Managed Agents in Gemini API: background tasks, remote MCP and more — Google DeepMind announced new capabilities for Managed Agents in the Gemini API, including background execution for async interactions, native remote MCP server integration, custom function calling, and credential refresh across sessions. 🔗 Graph: Google, Agentic AI, Model Context Protocol, Model Agnosticism 📅 Published: 2026-07-07 📰 https://blog.google/innovation-and-ai/technology/developers-tools/expanding-managed-agents-gemini-api/ 📌 Key takeaways: • Background execution lets developers pass background: true to run agents asynchronously, returning an ID for polling status or reconnecting later — eliminating fragile long-held HTTP connections. • Native remote MCP server integration means agents can access private databases and internal APIs without custom proxy middleware, mixing remote tools with built-in sandbox capabilities like Google Search and code execution. • The Updates also include custom function calling and credential refresh, addressing key developer feedback for production-ready agent deployments. • For Brett's stack: This directly validates MCP as an emerging standard for agent-tool connectivity — the same protocol used in TritonAI's agent architecture and Henry's infrastructure.

• Brown Professor Suspects Majority of His Class Used AI to Cheat — Brown University economics professor Roberto Serrano found that dozens of students in his spring course likely used AI to cheat on a take-home midterm, leading to an in-person final that caused mass drops and failures — and raising hard questions about institutional response to AI-enabled cheating at scale. 🔗 Graph: Higher Ed AI, AI Governance, AI Security 📅 Published: 2026-07-08 📰 https://www.insidehighered.com/news/faculty/learning-assessment/2026/07/08/brown-professor-suspects-most-his-class-used-ai-cheat 📌 Key takeaways: • Serrano gave a take-home midterm after student anxiety from a December mass shooting on campus; class enrollment ballooned from ~30 to 86, which he attributes to the expected take-home format. • The average midterm score was 96%, compared to a historical range of 65–80%; running the test through ChatGPT produced answers closely mirroring student submissions. • Serrano changed the final to in-person; more than a dozen students dropped the course and even more failed, suggesting many had relied on AI throughout the semester. • The professor characterized the university administration's response as "meek," underscoring the gap between faculty enforcement and institutional policy on academic integrity in the age of AI. • For higher-ed IT leaders like Brett: This case is a real-world stress test of AI governance policies — detection alone isn't a strategy; institutions need assessment redesign alongside AI adoption.

• Australian Payments Plus moves faster with ChatGPT and Codex — Australian Payments Plus (AP+) deployed ChatGPT Enterprise and Codex across its payments infrastructure organization, reporting dramatic efficiency gains: 77% of employees save 2+ hours per week, and complex reconciliation that previously took 4 hours now takes 30 minutes with Codex. 🔗 Graph: OpenAI, Codex, AI Adoption 📅 Published: 2026-07-07 📰 https://openai.com/index/australian-payments-plus 📌 Key takeaways: • AP+ operates Australia's payments and identity infrastructure, supporting products used by millions daily — a high-stakes environment where speed and accuracy both matter. • Codex reduced complex reconciliation investigation time from 4 hours to 30 minutes, and building working simulations dropped from days/weeks to 1 day. • 80% of surveyed employees reported improved creativity or work quality; the company's Chief People Officer emphasized the goal is "not simply greater efficiency" but helping people do their best work. • For UCSD's TritonAI program: This case study demonstrates concrete ROI for AI adoption in complex institutional workflows — directly analogous to the kinds of efficiency gains Brett is targeting with the Developer API Program and TritonAI Harness.

• Run AI workloads on any cloud, store on Hugging Face: zero-egress storage with SkyPilot — Hugging Face and SkyPilot announced zero-egress cloud storage integration, allowing AI workloads to run on any cloud provider (AWS, GCP, Azure, OCI) while storing models and datasets directly on Hugging Face, eliminating costly data transfer fees and vendor lock-in. 🔗 Graph: Hugging Face, AWS, Google Cloud, Model Agnosticism 📅 Published: 2026-07-07 📰 https://huggingface.co/blog/skypilot-hf-storage 📌 Key takeaways: • The integration lets users run training and inference workloads across any major cloud provider while storing artifacts on Hugging Face, with zero egress charges between SkyPilot-managed compute and HF storage. • This directly addresses the multi-cloud portability problem — a core tenet of model agnosticism, which is a top strategic principle in Brett's TritonAI architecture. • For UC San Diego's hybrid infrastructure strategy (on-prem at SDSC + cloud burst), this eliminates a key friction point: data gravity lock-in with individual cloud providers. • The integration also simplifies MLOps by decoupling compute from storage — teams can provision the cheapest or most available GPU region without migrating data first.

💡 Signal: This week's strongest signal is the convergence of agentic infrastructure maturity — Google shipping managed agents with native MCP support (legitimizing the protocol), OpenAI showing Codex delivering real enterprise ROI, and the entire industry (OpenAI, Anthropic, Microsoft, Amazon) collectively funding workforce transition through Raise US. For Brett, the takeaway is that agentic AI is moving from experimental to production-ready, the MCP bet appears prescient, and the workforce transition question is becoming a first-order strategic issue that higher-ed leaders need to engage with now, not later.