Design and implement a complete AI learning and research workflow for your field.
Module · Career Building
Lesson 12 of 12 available lessons
⬡ What you'll build
You've learned the pieces: how to decide when AI helps learning, how to prompt well, how to verify output, how to run a research pipeline, how to write with AI as a partner, and how to position and prove your capability. This capstone connects them into one system you'll actually use every week — and, documented well, it doubles as the strongest artifact in your portfolio. This is where the track stops being lessons and becomes your operating system.
Design, build, and run a personal AI-integrated study system: a documented, repeatable set of workflows that you use for real coursework over 2–4 weeks, then review and improve.
"System" is the operative word. The goal is not a pile of clever prompts or a one-time experiment. It's a connected set of habits and templates you can hand your future self at the start of any module and have it just work. It should be specific to your field and your actual courses — a generic system helps nobody, least of all you.
Six components, each drawing on a lesson you've already completed. Together they cover the full loop from encountering material to demonstrating capability.
| Component | Draws on | What it produces |
|---|---|---|
| Learning workflow | Learning Acceleration module | How you use AI to understand material without skipping the thinking |
| Prompting habits | Prompting Fundamentals | Your reusable prompts for explain / quiz / critique |
| Research pipeline | AI-Assisted Research Workflow | Scope → discover → retrieve → read → synthesise, with sources captured |
| Writing + verification | Writing with AI / Verifying Output | How you draft, check, and keep authorship + citation integrity |
| Capture | Note-Taking + Portfolio | Where notes, sources, and project write-ups live |
| Positioning output | Career + Portfolio lessons | At least one documented artifact that proves the system works |
If a component doesn't trace back to something you learned in the track, you're overbuilding. If a lesson's skill appears nowhere in your system, you're leaving value on the table.
You can structure this however suits you, but here is a clean default. Think of it as three layers: a home, a set of repeatable workflows, and an output.
The home — one place everything lives
The learning loop — for each new topic
The research pipeline — for each assignment
The writing + verification pass — for each output
The capture step — after each piece of work
The output — one documented artifact
A reusable prompt sits at the centre of the learning loop. Store a version like this in your prompt library and adapt it per subject:
I'm learning [topic] for [course]. I've just read/attempted the material myself — here's my understanding in my own words: [your attempt]. Do three things: (1) point out where my understanding is wrong or incomplete, citing the specific concept; (2) ask me five questions that test whether I really understand it, hardest last; (3) do not give me the answers until I respond. Add nothing that isn't grounded in standard material for this topic.
Your system is finished — for a first version — when it meets every row below. Score yourself honestly; a system that only half-exists won't survive a busy week.
| Criterion | Not done | Done |
|---|---|---|
| Reusable | Notes scattered across apps | One home you open by default |
| Covers the loop | Only one or two components | All six components present |
| Field-specific | Generic advice | Tailored to your actual courses |
| Verification built in | Checking is an afterthought | A verification step is part of every output |
| Honest AI use | AI does the thinking | AI accelerates; you retain the learning |
| Documented | Exists only in your head | Written up so you (or a reviewer) can follow it |
| Actually used | Designed but untouched | Run on real coursework for 2+ weeks |
There's no single correct shape. Real, valid capstones from different fields:
What they share: a home, the six components, honest AI use, and a written-up output. What they don't share: the specifics — because the system is yours.
A system you designed but never stress-tested isn't done — it's a hypothesis. The final, non-optional step is to use it on real coursework for two to four weeks, then run a short review.
Use it for real, for 2–4 weeks
Log the friction
Review and revise
That review closes the loop the whole track has been teaching: attempt, check, verify, and improve based on what actually happened — applied now to your own system. A student who can show this — a working system, used for real, honestly reviewed and improved — has demonstrated exactly the judgement, verification, and ownership that make AI a multiplier instead of a crutch.
Building a Portfolio That Shows AI Capability
Document this system as your first, strongest portfolio project using the five-section template.
Building Your Personal AI Learning Stack
The tools-and-workflows foundation this capstone assembles into a full system.
AI-Assisted Research Workflow for Students
The research pipeline that forms one of your system's six components.