LLM109 · Final Project: An AI Assistant for a Real Business Process
About Course
Design, build, evaluate and document an AI assistant that improves one real business process, bringing together everything from the program.
Who it is for
Learners who have completed the LLM program courses and want a substantial, portfolio-ready project.
Prerequisites
LLM101 through LLM108, or equivalent experience building, evaluating and deploying LLM applications.
What you will be able to do
- Scope an AI assistant around a measurable business process problem with clear users, inputs and success criteria
- Build a working assistant that combines structured outputs with retrieval or tools as the problem requires
- Evaluate the assistant against a baseline with a labeled test set and documented guardrails
- Present the solution with architecture, results, costs, risks and limitations in a professional case study
Lessons
- Lesson 1 · Project brief and scope
- Lesson 2 · Milestones and method
- Lesson 3 · Quality checklist and common mistakes
- Lesson 4 · Deliverables, rubric and submission
Lab: Project build environment
You will set up a single repository that holds your code, prompts, evaluation set, documentation and case study. Every milestone is committed so the history shows how the project evolved, and the final version runs from documented commands on a clean machine.
Free tools: VS Code or Google Colab, GitHub free, Docker Desktop, promptfoo, and the Gemini API free tier or Ollama with open models
Assessed assignment
Choose one real business process, such as answering supplier questions for a restaurant, triaging maintenance requests for a property manager, or preparing first-draft quotes for a print shop. Build an AI assistant that measurably improves it, compare it against a simple baseline on a labeled test set of at least 40 cases, and add the guardrails and human review the risks require. Deliver the repository, a demo recording and a case study of 1,500 to 2,500 words explaining the problem, design, results, costs, risks and next steps.
| Criterion | Weight | What excellent work looks like |
|---|---|---|
| Problem framing and scope | 20% | A specific process and user, a measurable success metric and a scope that is realistic for the time available. |
| Technical implementation | 30% | The assistant works end to end, uses structured outputs and retrieval or tools appropriately, and is packaged to run from documented commands. |
| Evaluation and safety | 30% | A labeled test set of 40 or more cases, comparison with a baseline, calibrated grading, guardrails and human review where needed. |
| Communication | 20% | A clear case study and demo that a business owner and a developer could both follow, with honest limitations and costs. |
Portfolio piece
A complete, portfolio-ready AI assistant project with repository, evaluation results, demo recording and case study.
Next step
You have reached the final course of this program. Keep growing with a Learning Path that combines this program with related ones.
Course Content
LLM109 · Final Project: An AI Assistant for a Real Business Process (8 h)
-
Lesson 1 · Project brief and scope
-
Lesson 2 · Milestones and method
-
Lesson 3 · Quality checklist and common mistakes
-
Lesson 4 · Deliverables, rubric and submission
-
LLM109 Assessment · Final Project: An AI Assistant for a Real Business Process