Project Overview
For this class’s individual project, you will design a process that uses agentic AI to solve a well-defined, meaningful task. Projects can use AI in a variety of ways: they can be particularly powerful AI agents, or more streamlined workflows leveraging AI, and we provide a few examples below. However, the goal is not only to build a working prototype, but also to demonstrate the value of leveraging AI in the particular setting you choose. Therefore, be ambitious and choose a topic you know well and love! We hope that whatever you create can have a real impact.
Application Examples
Your goal is to propose a use of agentic AI that would positively impact a company/organization you are deeply familiar with, or support a startup/project idea you have. Here are a few examples of topics:
- Daily interactive summaries of news coverage that are tailor-made to be relevant to an investment company’s portfolio, helping them with investment decisions.
- The automation of editing a hospital’s patient medical records based on recordings of doctor/patient interactions.
- A personal email agent that can summarize your emails for the day. It also decides which emails it will write draft responses for, ignoring things like subscriptions and blogs. This agent makes a list of “to-do” items for you and follows up after n days if something is not completed.
- Helping students choose courses to bid for at Kellogg given their career goals, as well as ratings of classes.
- Coordinating consultant expenses on a large consulting project, including consolidating receipts and credit card charges, categorizing expenses, flagging over-budget and non-compliant items, and updating internal budget trackers.
- Using medical records and police reports to write formal letters of demand for an insurance company.
Deliverable
You will create a pitch video addressed to the stakeholders of the organization you want to help, or to potential investors of the startup if you’re creating something new.
This project is the main deliverable of the course. It will be due at the same time as the final exam (end of exam week). It should include:
- An 8-minute pitch video, which should contain:
- An explanation of the problem you want to solve.
- A description of how you plan to leverage AI for it.
- A live demonstration of a working prototype (MVP) of your solution, which must include a use of n8n.
- An evaluation process MVP, with a live demonstration.
- A convincing explanation of why your process would be useful in practice, with an implementation path and discussion of risks and change management.
- A working n8n workflow and evaluation pipeline that you presented in the video.
Rubric
The project is by a large margin the most important component of your final grade (see the Syllabus page). Here is how it will be graded:
| Category | Description | Weight |
|---|---|---|
| Potential for impact | AI is leveraged to help with a well-defined, meaningful task that you know well and have experience with. The scope is appropriate for the course timeline. You convince us that it’s not only about “using AI” but that it would truly be useful. You discuss not only the AI process itself but how it would integrate into the broader organization you are targeting. The originality of the idea is also part of this grade. | 40% |
| Prototype | You are able to showcase and share a working prototype. This prototype should use n8n and an AI API (but not necessarily only that). It should also have a working evaluation pipeline in n8n. Complexity is not needed to get a high grade, but it should show that you seriously thought about the approach and iterated on it. The fact that the user experience looks nice is a plus. The teaching team should be able to run the n8n part of your prototype. The prototype does not have to be a fully working version of your idea; think of it as a minimal viable product. | 40% |
| Presentation | How engaging and creative your pitch video is. | 20% |
Timeline/Milestones
The course is short! It’s important to start early, and Kai homework will be there to help you stay on track.
| Week | Goal |
|---|---|
| 1 | Rough project idea, including features, target audience, and the problem being solved through this product. |
| 2 | Start to use n8n to implement your idea and iterate. Additional features that may be needed are identified. |
| 3 | Keep iterating. |
| 4 | Work on your evaluation pipeline, keep iterating on the project. |
| 5 | Think more deeply about the project implementation, the road to real impact, and the associated risks/change management. Keep iterating. |
| Exam week | Iterate on the video and how you choose to present the project. It’s not easy to condense content into a clear 8-minute pitch. |
Getting Help
While the project is individual, there are many ways to get support:
- The recitations will teach you n8n and other tools that could be useful.
- In-person office hours with the professor and TA will be held every Wednesday.
- Some of the course homework will be used as project check-ins.
- Kai (our AI assistant) is specifically trained to give you feedback and help you along the way if you. have any doubt.
- Peer help is allowed and encouraged – consider using the course’s Slack channel for it.
Advice and Rules
- Less is more! A typical bias is to try to build a complex agent or AI workflow. But in practice, the best projects are often the simplest, as they are easier to implement, improve, and evaluate.
- A consequence is that finding a great application is more important than creating a great AI system.
- Focus on a problem you deeply know. It is so much easier to know how AI could be integrated if you’ve experienced the situation yourself. And your video will be much more convincing and easy to create!
- Don’t hesitate to aim for an actual implementation after the course! This is not required at all, but ambition makes things more exciting, and being in touch with actual stakeholders would definitely boost your project and motivation.
- Start early! 5 weeks is very short, as great projects require iterations and sleeping on your ideas several times.
- You can work with several people on the same project, with restrictions.
- The deliverable should be individual to each student and will be graded separately. That is, each member must record their own video.
- No more than 3 participants per project.
- In the video participant must present a different part of the prototype that they worked on themselves. Therefore, each participant should have a different n8n workflow/evaluation MVP.
- For example, if Alice and Bob work together on a project, they could each focus on a different part of the prototype and evaluation system. Both Alice and Bob should make a separate 8-minute video, where they present the project in their own words and focus the demo on the parts that they worked on. The expected total amount of work for their project should be twice that of an individual project.
- The video should last 8 minutes. For full grade, the video should be in the 7min-8min15s range.