Customer stories

CODE University halved feedback check-ins in some modules. With AI agents its professors built on Langdock.

  • Emma Kudlich, Langdock
In short

CODE University of Applied Sciences in Berlin teaches software engineering, design and business management through project-based learning. Langdock has been in use across the university since 2023. Professors built agents for their modules that guide students with questions and give feedback before assessment. In some modules, the number of pre-assessment check-ins dropped by half. Faculty also use agents to prepare thesis assessments and oral exams, and students use Langdock in their own projects.

  • 50%fewer pre-assessment check-ins in some modules
  • 2023university-wide rollout of Langdock

The starting point. More students asking for feedback, the same resources.

As CODE grew, more students asked for personalized feedback and guidance, while the university's resources stayed the same. Faculty prepare digital classrooms, assess projects with many parts and give each student feedback before assessment.

Faculty also had doubts about AI in teaching. AI had to support students' critical thinking and keep them engaged with their own work, not replace that thinking.

Langdock's founders studied at CODE. They gave subscriptions to CODE students and faculty, and Langdock became the place where both groups tried out AI.

The solution. Module agents that ask questions instead of giving answers.

Since 2023, Langdock has been in use across the university, in three ways.

Feedback agents for modules. Professors built agents for specific modules, such as product strategy and decision-making, and shared them with students. The agents guide students through methods by asking reflective questions instead of giving direct answers, like a coach with a coachee.

From module agents to one meta-agent

The faculty later combined the module agents into one meta-agent. It draws on instructional materials, assessment criteria and task guidelines, and it tells the difference between guiding a student through a method and giving feedback on an assessment.

Preparing classes and assessments. Agents summarize interview data from bachelor theses, flag inconsistencies and draft questions for oral exams.

Students' own work. Students use Langdock for literature reviews, explanations of complex algorithms, code optimization and debugging, and project management from ideation to execution.

“The agent gives students immediate, high-quality feedback similar to what they'd receive in a one-on-one session. It has not only reduced my workload but also improved the depth and quality of their submissions.”

Florian GroteProfessor of Product Management, CODE University

The outcome. Half the pre-assessment check-ins in some modules, more time for mentoring.

In some modules, pre-assessment check-ins dropped by half, and some students no longer needed them at all. Faculty saved hours of preparation and spent them on mentoring and curriculum design.

The agents give more thorough and consistent feedback than professors could in the time available, and the quality of student submissions improved. Summarizing long interview transcripts with AI made thesis evaluations faster and more accurate.

With AI built into the curriculum, students learn to engage with it critically.

“Langdock has fundamentally changed how we teach and assess. It's not just about automating tasks, it's about creating a richer, more dynamic educational experience.”

Florian GroteProfessor of Product Management, CODE University
CODE University Challenge
Source: CODE University
CODE University Challenge
Source: CODE University
CODE University Challenge
Source: CODE University

About Code University

CODE University of Applied Sciences is a university in Berlin, Germany, with programs in software engineering, design and business management, taught through project-based learning.

Industry
Public sector, education & non-profit
Team
Company-wide
Products
Agents
Emma KudlichBrand & Marketing, Langdock

Written January 2026