IllinoisChat
An educational AI platform that answers questions grounded in your own course and research material — with citations, document ingestion, and course-level access control. No coding required.
What it is
IllinoisChat (also at chat.illinois.edu) is an educational AI web app from NCSA/Illinois for students and faculty. Its core is retrieval-augmented generation (RAG): you upload course materials, and the assistant answers questions from those documents, citing the passages it drew from.
It ingests PDFs, DOCX, PPTX, crawled web pages, Canvas LMS content, and PubMed, and organizes everything by course so each class has its own private corpus. It also powers several domain corpora such as patents, agriculture (“CropWizard”), and PubMed.
When to use it
Good fit
- You’re an instructor building a course Q&A assistant.
- You want answers grounded in specific documents, with citations.
- You need a web app, not an API or code.
- You want per-course access control for students.
Look elsewhere if
- You need raw programmatic model access — use Lumen or LLM Hub.
- You want to run open models yourself on HPC — see LLM Hub.
- You need large offline batch jobs — see LLMFlux.
Key capabilities
| Models | Answers can come from major commercial models (OpenAI, Anthropic, Google, and others), NCSA-hosted models, or a private model that runs entirely in your browser (nothing leaves your device). You choose the model per conversation. |
|---|---|
| Documents | Upload PDFs, Word, and PowerPoint files, crawl web pages, and pull in Canvas course content and PubMed — the assistant answers from them with citations. |
| Access | Sign in with your Illinois account. Instructors control who can join each course — owner, admins, and approved members — and API keys are available for programmatic use. |
| Cost | Free for the campus community. Usage is tracked per course so instructors can see activity, but there are no charges or caps for users. |
Data caution: many models are commercial and external
IllinoisChat can route to NCSA-hosted models and even run in your browser with WebLLM, but by default it also routes to commercial providers such as OpenAI, Anthropic, and Google. When you pick one of those, your prompt and any uploaded content are sent to that vendor. For sensitive course material, choose an NCSA-hosted or WebLLM model, and never upload regulated student data (FERPA) to an external provider without an approved agreement.
What’s recorded: IllinoisChat logs conversations. You can remove chats from your own history, but a copy is retained for course monitoring and quality analysis, and conversation content is also sent to a third-party logging service. Both web and API interactions are logged, so treat IllinoisChat conversations as recorded and avoid entering anything you wouldn’t want kept.