Capstone Project
The point of learning 26 tools is knowing how to chain them. Today you prove you can.
Everything you have learned over the last 28 days sits in your head now as individual tools. The skill that actually matters — the skill that distinguishes someone who knows AI from someone who is using AI — is knowing which tool to reach for at which moment, and how to chain them so the output of one feeds cleanly into the next. That skill cannot be taught. It can only be practiced. Today is the practice.
Why this tool matters
The capstone is not a finale; it is the exam. Not an exam of recall, but of judgment. You have 26 tools available. Your job is to pick a real project you need to ship, identify where each tool fits in the workflow, execute the chain, and ship the result. The judgment — which tool, which moment, how to hand off output from one to the next — is the skill the whole course has been building toward.
The specific project does not matter. What matters is that it is real: a presentation you actually need for Friday, a research memo your work actually wants, a video for a channel you actually post to, a mini-app for a problem you actually have. A made-up capstone produces a made-up workflow. A real capstone produces a workflow you will use again next week.
The one rule: use at least three distinct tools from the course, and design the chain so that the output of one tool becomes the input of the next. This constraint forces you past the “one tool, one task” habit and into the compose habit that is what AI-fluent professionals actually do every day. The difference between “I use AI” and “I work with AI” is the ability to describe your workflow as a sequence of deliberate handoffs.
Setup
Pick the project first. Before you open any tool, spend 10 minutes deciding what to build. Good capstones have: (1) a real audience or use, (2) a clear deliverable with a shape (a deck, a paper, a video, a mini-app, a dashboard, a literature review), and (3) a deadline within the next two weeks. Bad capstones are open-ended explorations without a shipping moment.
Common capstone shapes from past cohorts:
- The Research Brief. A 2-page memo on a question you actually care about, chaining Perplexity Deep Research + NotebookLM + Claude Projects + Gamma for the final deck.
- The Explainer Video. A 3-minute explainer on a concept from your field, chaining Claude Projects (script) + ElevenLabs (narration) + Runway (b-roll) + Descript (edit) + Captions (final mobile cut).
- The Social-Repurposing Pipeline. Turn one existing long-form piece (a lecture, a podcast, an article) into a month of social content, chaining Otter.ai + Opus Clip + Pictory + Captions.
- The Student Research Notebook. A literature-review notebook on a thesis-worthy question, chaining Elicit + SciSpace + QuillBot (citations) + Gamma (final presentation).
- The Mini-App Ship. A small personal tool deployed to a live URL, chaining Claude Projects (planning) + Bolt.new (build) + v0 (UI polish) + Cursor (custom logic).
Walkthrough
Step 1: Decide the shape of the deliverable
Write one sentence describing what you will produce by the end. “A 5-minute explainer video about why compound interest is the single most important concept in personal finance, posted to YouTube by Sunday.” Make the sentence concrete enough that a stranger could tell you when you're done.
Step 2: Map the workflow on paper
Sketch the workflow as a chain: Claude Projects (script) → QuillBot (tighten prose) → ElevenLabs (narration) → Runway (b-roll generation) → Descript (edit + captions) → Captions (final vertical cut). Each arrow is a handoff: what goes from one tool to the next, in what format?
Step 3: Build an asset bookkeeping doc
In a Google Doc or Notion page, track every artifact: the script, the narration audio file, the generated video clips, the editing project. Name files consistently. In a multi-tool workflow, 40% of the friction is finding last step's output. Solve it upfront.
Step 4: Work through each tool in sequence
Resist the urge to multitask. Do one tool's portion fully before moving to the next. Review the output. Fix or regenerate before handing off. A clean input to each tool produces a clean output; a messy hand-off compounds into disaster three steps later.
Step 5: Ship it, even if imperfect
At some point you will be tempted to regenerate the narration one more time, re-edit the video once more, tweak the deck just a little. Set a cutoff. Ship at 80%. The capstone's value is the completion, not the polish. You learn more from shipping a rough one than from never shipping a perfect one.
Step 6: Write a 300-word project retrospective
After shipping, write a retro: which tools pulled their weight, which underdelivered, which handoff was the trickiest, what you will do differently next time. This retro is the single most valuable artifact of the capstone — it is the template for every future multi-tool project you will run.
Your turn
Basic Capstone: A ship-it-this-week project
Pick a small but real deliverable you need to ship this week. Chain exactly three tools from the course. Ship it to its intended audience (even if the audience is just yourself). Write a 200-word retrospective.
Examples: a team update deck (Claude Projects + Gamma + Grammarly); a 90-second social video from your latest blog post (Pictory + Descript + Captions); a personalized outreach email batch (Clay + Grammarly + your email client).
Goal: prove you can run a real chain end-to-end. Time matters less than completion.
Advanced Capstone: A project you'd put in your portfolio
Pick a capstone you would genuinely show to a prospective client, employer, or collaborator. Chain 5–7 tools. Ship to a real audience where feedback will follow.
Example shape 1 (for researchers): an accessible-audience summary of your research, delivered as a short explainer video with cited sources. Chain Elicit + SciSpace + Claude Projects + ElevenLabs + Runway + Descript + Captions + Opus Clip (for social shorts).
Example shape 2 (for creators): launch a piece of content across three platforms simultaneously. Chain a single long-form recording through Adobe Podcast + Descript + Opus Clip + Pictory + Captions, producing one YouTube video, three TikToks/Reels, and one LinkedIn post.
Example shape 3 (for educators): a single-topic micro-course across 5 days. Chain Gamma + HeyGen + Descript + ElevenLabs + QuillBot + Grammarly, producing a welcome video, four lesson modules with AI narration, and a written summary.
Ship it. Solicit feedback from three real people. Write a 500-word retrospective analyzing which tools compounded each other and which were redundant in your particular chain. This retro is your personal operating manual for the next year of AI-fluent work.
Pitfalls and pro tips
Tool collection without tool composition. The trap is using five tools in parallel instead of in sequence. If your “chain” is actually five separate workflows that happen to be in the same project, you have not done the real exercise. A chain is when the output of one step is the input of the next.
Scope creep midway. As you work, each tool suggests adjacent opportunities (I could also generate an intro bed in Suno, I could add an AI avatar intro with HeyGen). Resist every one that isn't in your original plan. Finishing a simple chain beats half-finishing an ambitious one.
Skipping the retrospective. The retrospective feels optional but is not. Most people finish a multi-tool project, feel relief, and move on — losing the specific insights about their own workflow. Those insights are what make the next project go twice as fast.
How it compares
The capstone format comes out of executive-education and engineering curricula, where a final integrative project is known to outperform a final exam on skill retention and professional application. Compared to more-structured capstone formats (with rubrics, submission portals, instructor review), this capstone is deliberately self-directed: you are accountable to your own intended audience, not to a grade. For serious learners this is more rigorous, not less — an audience that wants the thing you shipped is a harder judge than a course rubric.
When to use — and when not to
Do the capstone as soon as you have one real project in mind. Many people delay it waiting for the “right” project; the right project is the first genuine one. Ship something small and real before you try something ambitious. The skill you are proving is integration, not scale.
Do not do the capstone as a throwaway exercise with no intended audience. Without a real reader, viewer, or user, the chain becomes academic and the judgment muscle stays weak. If you cannot think of a real project, interview a friend about their work and build a deliverable they actually need.