Recordings and decks
AI Playground Tutorial
A 15-minute screen recording by University IT of the Stanford AI Playground at aiplayground.stanford.edu, the free Stanford-run workspace where anyone with a S…
Private: Stanford sign-in requiredA Working Introduction: Gemini Enterprise
University IT's 25-slide walkthrough of Gemini Enterprise, Google's AI platform for Stanford, prepared July 2026. It explains what Gemini Enterprise is and how…
Private: Stanford sign-in requiredGemini Enterprise Tutorial
A 47-minute screen recording by University IT walking through Gemini Enterprise as a Stanford user sees it, following the Working Introduction deck: signing in…
Private: Stanford sign-in requiredStanford AI API gateway Practical Guide
Hands-on guide to the Stanford AI API Gateway (aiapi-prod.stanford.edu), the University IT service that routes requests to OpenAI, Anthropic Claude, Google Gem…
Private: Stanford sign-in requiredA Working Introduction: ChatGPT Edu
University IT's 25-slide walkthrough of ChatGPT Edu, Stanford's secure ChatGPT workspace, prepared July 2026. It explains what ChatGPT Edu is and which data it…
Private: Stanford sign-in requiredChatGPT Edu Tutorial
A 19-minute screen recording by University IT of the ChatGPT Edu workspace as a new Stanford user sees it: signing in, the Standard plan and what it includes,…
Private: Stanford sign-in requiredPrograms
AI at Stanford overview
University IT's front door: every approved AI service by vendor, free tiers versus paid upgrades, Slack AI and Zoom AI Companion, links to the AI classes, and the responsible-use guides.
ChatGPT Edu
Standard tier free for active faculty, staff, students and postdocs with a SUNet ID (the same models as OpenAI's $20 Enterprise plan); Pro needs a PTA. Approved for high-risk data and PHI. Request an account from the Get started links.
Claude for Education
Claude Chat, Claude Code and Cowork. Standard tier free with a SUNet ID, Premium (about five times the quota) needs a PTA. Chat is approved for all data except PHI; Code and Cowork up to high-risk. Sign in at claude.ai with SSO.
Gemini Enterprise AI
Agents, Deep Research and Gemini Notebook. Standard is free with a Stanford Google account and needs no request (40 queries a day); Pro needs a PTA (200 a day, no-code Agent Designer). Approved for high-risk data and PHI.
AI Playground
Free Stanford-run workspace at aiplayground.stanford.edu with OpenAI, Google and Anthropic models side by side, for all faculty, staff and students. Approved for high-risk data but not PHI.
AI Playground quick start guide
Signing in, choosing a model, plugins and assistants, installing the Playground as an app, and FAQs on tokens, context limits and hallucinations.
AI API Gateway
OpenAI-compatible API access to the Playground models for your own tools. Needs a PTA, approved for high-risk data and PHI, billed against a monthly budget you set. Request a key with the Add AI API Gateway Key form (two to five days).
AI API Gateway access and usage (KB00019978)
The real how-to, SUNet sign-in: base URL aiapi-prod.stanford.edu/v1, the Bearer key header, the model list (GPT, Claude, Gemini, DeepSeek, Llama) and the chat, embeddings and image endpoints with cURL and Postman examples.
AI code assistants with the API Gateway
Setup guides for Claude Code (CLI and VS Code), OpenAI Codex CLI and Cline against the gateway, connecting a project to GitLab, best practices per tool and example use cases; the route that lets Claude Code work with PHI.
Claude Cowork through the API Gateway
Five steps to point the Claude desktop app at the gateway (Developer mode, base URL aiapi-prod.stanford.edu, static API key, Bearer). Opus models recommended.
Responsible AI at Stanford
Be aware, be careful, be transparent: what never to paste into a third-party tool (moderate and high-risk data, PHI, passwords, source code, NDA material), why hallucinated citations happen, and best practices by area.
Responsible Agentic AI
For agents that act on your behalf: least privilege, read-only unless writes are essential, separate service accounts, human approval for consequential actions, and the risk tier every agent needs before deployment.
Approved AI services by data risk
Searchable matrix of every Stanford-provided AI service: provider, who can use it, approved risk level from low to high-risk with PHI, use case, and free or paid. Check it before any Stanford data goes into a tool.
ChatGPT Edu on your phone (KB00020806)
SUNet sign-in. Install the official OpenAI app, sign in with Stanford SSO, pick the Stanford Edu workspace, then work by text, voice and photo; chats sync with web and desktop.
Claude Chat Projects (KB00020839)
SUNet sign-in. Set up a Chat Project in claude.ai so standing instructions and reference files (up to 30 MB each) follow every conversation, and when a Cowork Project is the better fit.
Slide decks from Gemini Notebook (KB00020815)
SUNet sign-in. Turn the sources in a notebook into a PPTX with speaker notes and citations, as a detailed deck or presenter slides, at aiworkspace.stanford.edu.
Move personal Claude chats to Stanford (KB00020984)
SUNet sign-in. Two best-effort ways to carry a personal-account conversation into your Stanford Enterprise account: export a reference file or ask Claude for a full summary, then upload it.
