Course Kickoff: Why AI Agents Now

Day 1 - Block 1

Finn-Ole Höner

PhD Candidate in Quantitative Marketing (ESE)

LLMs for Innovation:

  • Ideation with AI Agent teams (Dr. van Crombrugge, Dr. Wichmann, Prof. Stremersch)
  • Refining and validating ideas for marketing texts (Prof. Donkers, Prof. Fok)
  • Synthetic respondents for sensitive topics (Prof. de Jong, Prof. Pieters)

Background:
BDS, Deutsche Bundesbank, Lidl, University of Tübingen

Why do we care?

A Nature survey of scientists found that …

{background-image=“../../assets/images/epoch_ai.png”, background-size=“contain”}

Q1 Programming AI use (n=25)

  • Course link: onboarding, LLM basics, verification loop

Q2 AI use in research

  • Course link: ideation, literature review, writing, analysis, data collection

Q3 Top priorities (free text)

  • Course link: tools, tool building, memory, rigorous analysis, workflows

Q4 Laptop mix

  • Course link: setup support and in-class troubleshooting

Q5 Paid subscriptions

  • Course link: local and open workflows; no paid-only requirements

Q6 Direct API experience

  • Course link: tool calls first, MCP builder second

Q7 Language profile

  • Course link: language-flexible workflows, tool-first demos, verification habits

Q8 Comments, wishes, questions

  • Requests: Stata fit, multi-agent practice, security guidance
  • Concern: restricted data leakage when using agents
  • Course link: privacy-by-design and permission-restricted modes

(A necessary) Disclaimer

  1. Massive amounts of work on this topic are happening right now
  2. Many, many (legal, ethical, environmental, social) questions are open
  3. AI is fuzzy + a lot of hype

Wide overview -> Experiment on your own!

Meet Jan: AI tutor for this course!

Github: finnoh/ti-student-agent-pack

You: I want to work on E03.

Jan: +------------------------------------------+
     | Exercise E03 - Context + Retrieval      |
     | Objective: Compare baseline vs retrieval|
     | Deliverable: A/B note + cited chunk     |
     +------------------------------------------+

     First step: run the baseline (no retrieval)
     and paste the output so we can compare fairly.

Installation:

curl -sL https://raw.githubusercontent.com/finnoh/ti-student-agent-pack/main/install.sh | bash

How to Use Jan

  1. Open a terminal
  2. cd path/to/student-agent-pack
  3. opencode

Hands-on: Explore Jan

Task: Set up Jan and have a look around

  1. Install Jan: curl -sL ... | bash (or use the course pack zip)
  2. Open student-agent-pack/
  3. Explore the directory structure and files (what do you see?)
  4. Chat with Jan (what does Jan do?)
  5. Can you customize Jan to your needs?

References

TI Workshop | D1 B1