What's New in Your AI Tools
Zoom AI Companion
- Zoom sets new migration dates for the new web experience — Zoom has published new migration dates for the redesigned Zoom web experience: September 30, 2026 for Enterprise (direct sales) and Release Track customers, and October 31, 2026 for indirect/channel accounts. Admins can move their organization over at any time before their segment's date, and can roll back to the previous experience until then. After the deadline, the new experience is enforced and rollback is no longer available. If your team runs a departmental Zoom account, worth checking which segment you're in before the fall.
Microsoft Copilot with Data Protection
- Copilot Chat now opens alongside Copilot Search — Copilot Chat is now available in the right-hand side pane while using Copilot Search in the Microsoft 365 Copilot app, so you can act on search results without switching pages or losing context. Also in this update: email references now open directly inside Copilot Chat instead of launching a separate Outlook window, and the Copilot app has a refreshed design with a simplified chat home screen and navigation pane. Small changes individually, but they cut down on the app-switching that makes AI features feel like extra work.
Copilot for Microsoft 365
- Outlook emails can now be added as references in Copilot Notebooks — Copilot Notebooks now accept Outlook emails as knowledge sources, alongside the files and documents already supported. That means the decisions and caveats that live in email threads can now ground a notebook's outputs directly, rather than being re-summarized by hand first. The notebooks interface also got a streamlined navigation update in the same release.
Google Gemini & NotebookLM
- Ask Gemini in Chat starts rolling out — Google is introducing Ask Gemini in Google Chat, a unified command line for your work powered by Workspace Intelligence, with gradual rollout starting August 26, 2026. Through October 1, Workspace customers get promotional access to higher usage limits, so it's a good window to try it before standard limits apply. If your team already works in Chat, this is the Gemini feature to test this quarter.
Coming Up: Trainings & Workshops
No live trainings are scheduled in the next 14 days, but you can:
- Take the self-paced AI Foundations course to learn core AI concepts and UC policies on AI tools.
- Watch the AI Webinar #6 recording — a practical walkthrough of TritonGPT's first year, including MyDocuments, model switching, and chat sharing.
- Explore the Everyday I AI video series for short, practical prompting tips that work across TritonGPT and other AI tools.
TritonAI News
- TritonAI publishes Citizen Developer Guide and updated data-protection guidance — TritonAI released a new Citizen Developer Guide to help staff, faculty, and researchers identify recurring work, choose a supported delivery path, and set data and human-review boundaries before building. The updated data-protection guidance clarifies that TritonGPT and the TritonAI Harness may be used with information up to P3 only within an approved service or setup, and that P4 data is not approved. If your team has been eyeing a workflow automation project, read this first.
- UC San Diego to Test Innovative Power Technology for AI Data Centers — The San Diego Supercomputer Center will serve as the testbed for an $8.48 million California Energy Commission-funded project to change how AI data centers receive and manage electricity, including a bidirectional solid-state transformer that converts medium-voltage AC directly to 800-volt DC. The project targets two megawatts of AI computing load, a 50%+ reduction in power equipment footprint, and roughly 25% projected energy savings, plus workforce pathways for under-resourced communities.
- NSF Funds New Center to Design Reliable, Self-Correcting Quantum Computers — A multidisciplinary team including UC San Diego computer scientists received a $37.5 million NSF grant for NSF PRACTIQAL, a center aimed at making quantum error correction practical. Yufei Ding, associate professor of computer science and engineering, leads a thrust on error-correction theory, scalable fault-tolerant architectures, and an open-source QEC software stack.
Have feedback on this newsletter? Get in touch.