AI That Teaches, Not Just Answers
Grounded in your course materials. Built into Artemis. Always ready.
30,000+ conversations powered at TU Munich
See Iris in Action
Four ways Iris supports students — from lecture Q&A to coding exercises and smart search.
Ask anything about a lecture. Iris searches your course materials, cites exact sources, and lets you jump straight to the relevant page.
Ask anything about a lecture. Iris searches your course materials, cites exact sources, and lets you jump straight to the relevant page.
Challenge yourself on any topic. Iris creates questions from your course content and explains every answer with direct source links.
Get guided help on assignments — in the browser or your local IDE. Iris sees your code, your progress, and nudges you toward the solution without giving it away.
Search across all course materials and get an AI-powered summary of the most relevant results — right from the search bar.
Why Generic AI Falls Short for Teaching
Generic chatbots give fluent, confident answers — but they don’t teach. Without course context or pedagogical design, they hand students the answer and skip the learning.
Student skims the answer. Forgets it tomorrow.
Student works through the problem. Genuine understanding.
Iris guides students to the answer instead of giving it away — research shows a significant boost in intrinsic motivation (Cohen’s d = 0.55).
| Feature | Generic AI Chatbot | Iris |
|---|---|---|
| Broad general knowledge beyond course scope | ✓ | ✓ |
| Knows your specific course content | ✗ | ✓ |
| Guides with hints instead of giving full answers | ✗ | ✓ |
| Cites lecture slides and course materials | ✗ | ✓ |
| GDPR compliant & university-hosted | ✗ | ✓ |
“If you need to do something fast and efficiently, you would use it. But if you do something just for learning, you would not.”
— P20, ChatGPT user
“I think it’s very easy to learn using ChatGPT. But next day I will forget because I just learned it from ChatGPT.”
— P20, ChatGPT user
Grounded in Your Course. Guided for Your Students.
Iris learns from lecture slides, transcripts, faqs, exercises, and student code — then provides context-aware hints inside Artemis.
Upload Your Materials
Add your lecture slides, transcripts, and FAQs to your course in Artemis. Click one button to send them to Iris.
Iris Learns Your Course
Iris reads every slide and transcript. It learns your terminology, your examples, and the way you teach the material.
Students Get Real Answers
When students ask questions, Iris responds using your course content — with citations back to specific slides.
Iris maintains context across conversations, adapting its hints based on prior interactions and student progress.
Used at Scale. Studied in Real Courses.
Iris is not a prototype. It has been deployed and evaluated in live teaching at the Technical University of Munich.
Key Research Findings
What Students Say
From interviews with 33 CS students in a controlled study comparing Iris, ChatGPT, and no-AI support.
“Iris was clearly aware of the context. It pointed me in the right direction. When I asked for getting the strings, it said, you can shift the strings like this for this algorithm without me even mentioning the algorithm.”
“Context awareness — like if ChatGPT would already know your code instead of you copy paste.”
“In my experience, I’ve seen that a part of your brain turns off when you are basically telling the AI to do the stuff and all of your focus is just diverted to see that the input and output is basically working or not.”
Quotes from Bassner, Lottner & Krusche (2025). “Towards Understanding the Impact of Context-Aware AI Tutors and General-Purpose AI Chatbots on Student Learning.” Koli Calling '25, ACM.
Designed for Institutional Trust
Governance, transparency, and course control are core product requirements — not afterthoughts.
Privacy by Design
Iris minimizes data shared with LLM providers. Only the context needed for a response is sent — no bulk uploads, no training on your data.
University-Hosted
Iris and Artemis run on your institution’s infrastructure. External cloud processing is limited to LLM inference — or eliminated entirely with local models.
Instructor Control
Instructors decide what Iris knows. You control which materials are indexed, and students can opt out of AI features entirely.
Choose Your Starting Point
Bring Course-Grounded AI to Your Teaching
Built into Artemis. Backed by peer-reviewed research at TU Munich. Talk to the team about a pilot, or explore the documentation.