Snowflake and Databricks govern AI agents as adoption grows

Key Takeaways

• Wilbur Wright College opened City Colleges of Chicago’s first AI degree; Northwestern launched an undergraduate AI major, and the University of Chicago plans an applied AI master’s for fall 2027.
• DePaul’s AI principles emphasize academic integrity, transparency and human-centered use; its lab studies AI’s effects on daily life.
• Former people chief Lynne Oldham says HR leaders need firsthand AI experience to guide workplace changes, after seeing investment fail to alter business operations.
• Recommended AI product-management skills for 2027 include agent design, evaluation, responsible AI, data literacy and prototyping; Pawel Huryn estimates AI-focused US product managers earn about $245,000 versus $123,000 for traditional roles.
• Snowflake offers agent permissions and auditing; Databricks uses Unity AI Gateway and Unity AI Catalog for monitoring and access controls. Dataiku, Oracle, Cloudera and Nvidia are also developing agent-management or deployment tools.
• One Identity Identity Manager 10.0 adds natural-language identity-data queries, risk integrations, behavior-based governance and custom threat-detection playbooks.
• Modal says asynchronous agent reinforcement learning can reduce weight-sync traffic 1,000-fold; cheaper rollout hardware cut costs 15% without destabilizing rewards.
• Qbim’s Q-Sales 2.0 processes emails, calls and meetings to update customer records, identify opportunities and recommend sales priorities.
• Twenty-one Penn State professors are testing Ray-Ban Meta smart glasses; the article does not specify how they will be used in classrooms.
• Video-aware AI could help customers perform some service tasks themselves, while MPR News plans a discussion of AI’s health-care uses and concerns on November 11.

Chicago City Colleges launches its first artificial intelligence degree

City Colleges of Chicago serves more than 73,000 students each year and is expanding its focus on AI. This fall, Wilbur Wright College introduced the system’s first AI degree, an Associate of Applied Science in Artificial Intelligence and Machine Learning Development. Provost Mark Potter said AI skills, including responsible and ethical use, matter across many fields. Associate Vice Chancellor Ayoka Noelle Mota Samuels said the colleges aim to make AI education affordable, trusted and open to people at different stages of learning.

Chicago universities expand AI programs and set rules for responsible use

DePaul University’s Responsible AI Systems and Experiences Lab studies how AI affects daily life, while a 26-member committee developed the university’s guiding principles for AI use. The principles call for academic integrity, transparency and human-centered use. Other Chicago schools are also expanding AI education: Northwestern University launched an undergraduate AI major this fall, and City Colleges of Chicago opened its first AI degree at Wilbur Wright College. The University of Chicago plans to launch a master’s degree in applied AI in fall 2027, while its law school bars electronic note-taking in some classes.

MPR News explores how AI is changing health care

MPR News host Angela Davis is scheduled to discuss AI in health care at 9 a.m. Wednesday, November 11. The program will look at how doctors use AI to diagnose diseases, treat patients and manage workloads, as well as concerns about accuracy, reliability and patient care. Listeners can call during the 9 a.m. hour with questions or share how they have seen AI used in health care. Guest Matt Redlon is chair of the AI program and vice president of digital biology.

HR leaders need hands-on AI experience to guide workplace changes

Lynne Oldham, a former chief people officer at Zoom, Stash and Dataiku, says HR leaders need firsthand experience with AI to guide changes to jobs, organizations and employee responsibilities. She recently returned to school to study AI strategy and learn to build with the technology. Oldham joined the HIC C-Level AI Cohort so her advice as a board member would draw on direct experience, not only presentations or vendor demonstrations. She has seen companies invest heavily in AI without changing how their businesses operate.

Modal says agent reinforcement learning is ready for production

Jason Mancuso of Modal said asynchronous agent reinforcement learning can reduce weight-sync traffic by 1,000 times. Modal’s approach starts with GRPO and uses sparse updates to move inference off the training cluster. Mancuso said agent RL needs about two inference GPUs for each training GPU, and Modal spreads rollouts across B200, H100 and A100 GPUs, as well as mixed H200 pools, using BF16 and FP8. He said cheaper rollout hardware cut costs by 15 percent without destabilizing rewards, and the RL system builds on Miles and SGLang, with full code expected to be released soon.

One Identity updates Identity Manager with AI governance tools

One Identity announced Identity Manager 10.0 in Aliso Viejo, California, on November 10, 2026. Its browser-based tools let administrators handle synchronization for connected systems and create, edit, and monitor automated schedules. Authorized users can ask questions about identity data in natural language, reducing the need for complex SQL. The update also adds risk-score integrations, behavior-based governance, and custom Identity Threat Detection and Response playbooks to help teams find and address identity risks.

Penn State professors test Meta smart glasses for education

Twenty-one Penn State professors are testing Ray-Ban Meta smart glasses as part of the 2026 Faculty Engagement Award. Penn State University Libraries’ Teaching and Learning with Technology department is hosting the program. The glasses were in use at the Huck Life Sciences Building in University Park, Pennsylvania, on October 29, 2026. The article raises the question of how the glasses could be used in education, but the provided information does not describe specific classroom uses.

Qbim launches AI sales platform Q-Sales 2.0

Technology company Qbim launched Q-Sales 2.0, an updated AI-driven sales system, along with a new website. The platform processes information from emails, phone calls, and customer meetings to keep customer records current, find business opportunities, and suggest which sales activities to prioritize. Qbim says it can reduce manual administration compared with traditional CRM systems and give sales teams more time to work with customers. CEO Niklas Forsberg said the launch supports the company’s goal of building an internationally scalable software business with a more product-driven sales model.

AI video tools could put some blue-collar service jobs at risk

AI assistants that understand video could help customers complete tasks themselves, including assembling flat-pack furniture. Using a phone camera, an assistant could identify parts, recommend tools, check work, and respond when someone gets stuck. The article says this could reduce paid jobs even if customers still do the physical work, though many will continue to hire workers for convenience or expert judgment. Service businesses are urged to review recent jobs and test camera assistance on low-risk tasks, measuring time, mistakes, and callbacks before changing their services.

Seven AI skills could help product managers stay essential in 2027

An article on AI product management says product managers must decide what AI agents should do, measure their results, and keep them safe as companies build more products around agents. It lists seven skills for 2027: probabilistic product thinking, AI evaluation and metrics, agent design and orchestration, responsible AI and compliance, machine learning and data literacy, stakeholder storytelling, and AI prototyping. Product coach Pawel Huryn estimates US AI-focused product managers earn about $245,000, compared with about $123,000 for traditional product managers. The article argues that training current staff may cost companies less than hiring AI-focused managers, while also pointing to risks such as safety, privacy, legal liability, and ethics.

Data Platform Companies Race to Control Enterprise AI Agents

As companies adopt AI agents, they face challenges tracking what the agents do across different systems and departments. Unlike chatbots, agents can take actions on their own, including accessing data and changing business applications. Dataiku’s Agent Management helps companies register and monitor agents, while Oracle’s AI Data Platform includes Agent Hub, which supports MCP and A2A standards. Snowflake offers individual agent permissions and auditing, and Databricks provides monitoring and access controls through Unity AI Gateway and Unity AI Catalog. Cloudera and Nvidia are also developing tools to help companies deploy agents with limits on what they can access.

Sources

• As AI stakes its claim in higher education, colleges and universities across Chicago aim to embrace it
• As AI stakes its claim in higher education, colleges and universities across Chicago aim to embrace it
• How artificial intelligence is changing health care
• AI Is Transforming the Workplace. HR Leaders May Not Be Ready to Lead It
• Modal says agent RL is ready for production
• News alert: One Identity launches Identity Manager 10.0 with AI-powered governance tools
• 21 Penn State professors are testing Meta’s AI smart glasses. What can they do for education?
• Qbim launches AI-based sales system Q-Sales 2.0
• Blue-Collar Jobs Aren’t as AI-Proof as They Look
• The AI Product Management Skills 2027 That Keep PMs Indispensable
• Who Controls Enterprise AI Agents? Data Platform Vendors Race to Provide the Answer