Capstone Project

Module 14 · Apply Everything You’ve Learned

Content current as of April 2026.

Learning Objectives

Builds on: All 13 modules — the capstone is the synthesis

By the end of this module, you will:

  1. Select a capstone option that matches your field, time budget, and career goals
  2. Produce a substantial deliverable that demonstrates integrated AI literacy
  3. Give and receive structured peer feedback using a criterion-focused protocol
  4. Present your work in a format that matches how AI work is shared professionally
  5. Self-assess against a 5-dimension rubric aligned with the AAC&U VALUE framework

You’ve Built the Foundation. Now Build Something Real.

Everything in the previous 13 modules pointed at one outcome: you can demonstrate AI literacy in a way that an employer, a graduate admissions committee, or a research lab will recognize. This module is where you prove it. Four capstone options are offered — each one valid, each one shippable in 2-4 weeks of focused work.

The rubric you’ll be assessed on is aligned with the AAC&U VALUE framework, used by 5,600+ higher education institutions across 159 countries. That alignment matters: your capstone score is a credential that travels with you.

This is not busywork. This is the artifact you will point to in your next interview.

Common Misconceptions About Capstones
  • “My project has to be perfect” — It doesn’t. It has to be real, tested, and honestly reflected upon. An honest project that failed and explains why is better than a polished project that was never tested.
  • “I need to build something technically advanced” — An Integration Plan that identifies the right AI tool for a real workflow is just as valuable as building a custom GPT. The capstone tests your JUDGMENT, not your coding ability.
  • “I should use as many AI tools as possible” — Quality over quantity. A project that uses one tool well beats a project that name-drops ten tools superficially.

The four capstone options

Read all four before you pick. The best option is the one that most closely matches work you’ll actually do in the next two years.

Option A — Integration Plan

What it is: A detailed plan to integrate AI responsibly into a specific workflow you already have — at work, at school, in a student organization, in a volunteer project. Not “someday.” Next month.

Deliverable: A 12–18 page plan covering:

  1. Current workflow map (with time and cost estimates)
  2. Identified AI integration points with rationale
  3. Specific tools selected with pricing
  4. Verification / quality-control protocol
  5. Risk assessment (privacy, ethics, failure modes)
  6. Pilot plan with success metrics
  7. Rollout timeline
  8. What could go wrong and what you’d do about it

Best for: Students who are already working part-time, running organizations, or have active projects. Also best for career-switchers who want to bring an AI-integration plan to a specific target employer as an interview artifact.

Example scope: “AI integration plan for the university writing center peer tutor scheduling system” — includes current pain points, integration points, tool selection, ethics review.

Option B — Custom AI Solution

What it is: A working AI system you build for a real user (not a demo). Could be a custom GPT, a Claude Project with skills, a retrieval-augmented workflow, a small agent, a specialized prompt library with verification scaffolding.

Deliverable: The working system plus a 10–15 page documentation package covering:

  1. Problem statement with user validation (who asked for this, what they need)
  2. Design decisions with trade-offs
  3. Technical approach and tools used
  4. Evaluation protocol (how you know it works)
  5. Known failure modes (what it gets wrong, how users are protected)
  6. Usage instructions
  7. Privacy and data handling
  8. Limitations and roadmap

Best for: Students with technical inclinations, or students who want a concrete portfolio artifact they can demo in an interview. Particularly strong for computer science, data science, engineering, and library/information science students.

Example scope: A Claude Project that helps undergraduate researchers draft methods sections for empirical papers in their specific subfield, with built-in citation verification and a two-pass accuracy check.

Option C — Workshop

What it is: A facilitated AI-literacy workshop you design and deliver to a real audience of 8+ people. Could be peers, community members, a student organization, a church group, a workplace team, a high school class.

Deliverable: Full workshop package:

  1. Learning objectives (3–5, measurable)
  2. Agenda with timing (90–120 minutes)
  3. Slide deck or visual materials
  4. Participant worksheet or workbook
  5. Pre- and post-surveys
  6. Facilitator notes
  7. A 5-page reflection on what you learned from delivering it, with anonymized participant feedback

Best for: Students heading into education, training, HR, social work, community health, library science, or any field where explaining AI to non-experts will be part of the job. Also strong for students who want to demonstrate leadership and communication, not just technical skill.

Example scope: “Prompting for Small Business: a 90-minute workshop for local entrepreneurs covering BRIEFS, verification, and ethical limits.”

Option D — Mini-Research

What it is: A small, original research project on an AI-related question. You formulate a specific question, gather or generate evidence, analyze it honestly, write it up.

Deliverable: A 10–20 page research note covering:

  1. Specific research question with motivation
  2. Literature / context review (5–10 sources)
  3. Method — how you generated or gathered evidence
  4. Results with raw data and analysis
  5. Discussion including limitations
  6. Conclusions with what you are and are not willing to claim
  7. Data and code in a public repository (or appendix for non-code work)

Best for: Students heading to graduate school, students curious about research careers, students who want to develop a writing sample that’s more rigorous than a term paper. This option develops the research skill set most directly.

Example scope: “A comparison of three prompting strategies on a set of 50 undergraduate statistics questions: accuracy, hallucination rate, and useful error types.” Or: “What do 20 syllabi from Fall 2025 biology courses say about AI use, and what does this imply for students entering the field?”

How to pick
  • If you already have an active project or job → Option A.
  • If you can code or build systems comfortably → Option B.
  • If you want to practice teaching and communication → Option C.
  • If you’re graduate-school-curious → Option D.

All four are equally respected. They’re just different shapes for demonstrating the same underlying competencies.

The planning worksheet

Before you start, fill this out. A capstone that skips planning usually fails in week 3, not week 1.

======================================================================
CAPSTONE PLANNING WORKSHEET
======================================================================

NAME: ________________________
DATE: ________________________

----------------------------------------------------------------------
1. OPTION SELECTED: [ ] A  [ ] B  [ ] C  [ ] D

2. ONE-SENTENCE PROJECT DESCRIPTION:
   (If you can't write this in one sentence, scope is too big.)

   ______________________________________________________________

3. WHO IS THE END USER / AUDIENCE?
   (Not "professors" or "classmates." A specific real person or group.)

   ______________________________________________________________

4. HOW WILL YOU KNOW YOU SUCCEEDED?
   (A specific measurable thing. "They use it" / "They learn it" /
    "The data show X" / "The plan is adopted." Not vague.)

   ______________________________________________________________

5. WHAT COULD GO WRONG?
   (List at least three real risks.)

   a. ______________________________________________________________
   b. ______________________________________________________________
   c. ______________________________________________________________

6. WHAT WILL YOU DO IF (5a) HAPPENS?

   ______________________________________________________________

7. TIME BUDGET — in hours, realistic:

   Week 1 (planning, scoping):         ______ hours
   Week 2 (building, drafting):        ______ hours
   Week 3 (iterating, testing):        ______ hours
   Week 4 (finalizing, presenting):    ______ hours
   TOTAL:                              ______ hours

   (If total is under 20 hours, scope is probably too small.
    If over 60 hours, scope is too big or you're perfectionist-
    scoping.)

8. WHAT HAPPENS TO THIS DELIVERABLE AFTER THE COURSE ENDS?
   (Portfolio URL? Published somewhere? Handed off to someone?
    A capstone that dies with the semester is half a capstone.)

   ______________________________________________________________

======================================================================

Fill this in. Get feedback on it from one peer before you move to week 2. The 30 minutes spent on the worksheet will save 3 hours later.

Milestone schedule (4-week default)

The default capstone runs four weeks. Adjust timing to your course calendar, but keep the structure: plan → build → iterate → finalize. Skipping the iteration week is the single most common failure mode.

Week Milestone Deliverable Checkpoint
Week 1 Planning complete Planning worksheet signed off Peer review: is scope realistic?
Week 2 Initial draft / prototype First full-length draft OR working prototype Self-check against rubric
Week 3 Iteration from feedback Revised draft with explicit changes logged Peer review using protocol (below)
Week 4 Final delivery Final deliverable + 5-min talk + peer reviews written for two classmates Submit

The week-2 self-check

At the end of week 2, before you continue, answer these five questions:

  1. Does the deliverable match the one-sentence description from your worksheet? (If not — which moved, the deliverable or the description?)
  2. Would the end user you named actually use / benefit from this? (If you haven’t asked them, ask them this week.)
  3. What is the weakest part of the current draft?
  4. What would it take to fix the weakest part in week 3?
  5. What is most likely to make you want to rewrite this from scratch, and how do you prevent that?

Question 5 is the most important. Capstone restart mid-week-3 is the #1 failure mode. Prevent it by answering question 5 honestly in week 2.

Presentation guidelines

Each capstone includes a 5-minute talk delivered to the class. This is non-negotiable and follows a specific structure.

The 5-minute talk structure

Time Section Content
0:00–0:45 The hook What’s the problem? Why does it matter? Who is affected?
0:45–1:30 What you built / planned / ran / found The one-sentence version. Show, don’t tell.
1:30–2:45 How you verified it works Your evaluation protocol. What did you measure?
2:45–3:45 What it gets wrong Honest failure modes. This is your credibility section.
3:45–4:30 What it would take to scale this / what you’d do differently The “if I had another month” or “if I did this at work” section.
4:30–5:00 Questions Take 1–2 questions. “I don’t know, but here’s how I’d find out” is an acceptable answer.

Talk rules

  1. No reading from a script. Have notes. Don’t read them.
  2. No more than 6 slides if you use slides at all. Many strong capstone talks use zero slides and one live demo.
  3. Cover failure modes in ≥45 seconds. Talks that skip this signal superficiality.
  4. Stop at 5 minutes. Timing discipline is part of the assessment.
What good looks like

A strong capstone talk ends with the audience thinking one of two things:

  1. “I want to see the full deliverable.”
  2. “I want to try this in my own work.”

If neither happens, something in the talk went wrong — usually either the hook was too abstract, or the failure-modes section was too short.

The peer review protocol

Peer assessment is validated research as an effective learning tool — when students are given structured criteria. Unstructured peer review (“what did you think?”) produces low-quality feedback. Structured peer review produces feedback often as reliable as instructor assessment.

The criterion-focused method

For each capstone review, you will:

  1. Score each of the five rubric dimensions on the 4-point scale (Emerging / Developing / Proficient / Exemplary) before writing any prose comments.
  2. For each dimension, cite one specific piece of evidence from the capstone (a line, a section, a claim) that justifies the score.
  3. Write one “continue” and one “consider”: one thing the capstone does well and should keep doing, one thing to consider changing.
  4. Identify the single most important question the reviewee should answer to strengthen the work.

This protocol is adapted from established research on peer assessment (Boud & Falchikov, Topping) and the AAC&U VALUE scoring protocols. The key finding across decades of research: the act of scoring someone else’s work against a rubric is itself a learning event for the reviewer, often as valuable as the feedback received.

What peer reviewers must avoid

  • Vague praise (“This is great!”) — not useful
  • Vague criticism (“This needs more work”) — not actionable
  • Judgment without evidence — violates the criterion-focused design
  • Comments outside the rubric’s scope — stay focused on what’s being assessed
  • Rewriting someone else’s work for them — you’re assessing, not producing

Timing

  • Week 3: each student submits draft; receives peer review from 2 assigned peers within 48 hours.
  • Week 4 (final): each student writes peer reviews for 2 other classmates’ final submissions.

The 5-dimension rubric

This rubric is built on AAC&U’s VALUE Rubrics. Each dimension maps to a specific VALUE rubric; the four performance levels correspond to VALUE’s Benchmark, Milestone 2, Milestone 3, and Capstone levels.

Why this alignment matters

The five-dimension rubric below is built on AAC&U’s VALUE Rubrics — used by over 5,600 higher-education organizations and accepted by all U.S. regional accrediting bodies as authentic assessment. Each dimension maps to a specific VALUE rubric:

  • Technical Competency ← VALUE Problem Solving + Inquiry and Analysis
  • Critical Thinking ← VALUE Critical Thinking
  • Ethical Reasoning ← VALUE Ethical Reasoning
  • Communication ← VALUE Written Communication (or Oral Communication)
  • Integration & Synthesis ← VALUE Integrative Learning

If you include this capstone in graduate school applications, tenure files, or portfolios reviewed by external funders, you can reference the AAC&U VALUE framework directly. That alignment is a credential in itself.

Source: AAC&U Valid Assessment of Learning in Undergraduate Education (VALUE) system. aacu.org/value

Dimension 1 — Technical Competency

(VALUE: Problem Solving + Inquiry and Analysis)

Level Descriptor
Exemplary (4) Uses AI tools and techniques skillfully. Demonstrates clear understanding of what the tools can and cannot do. Verification and quality-control protocols are explicit and appropriate.
Proficient (3) Uses AI tools appropriately. Some verification protocol present. Minor technical gaps, but nothing that undermines the work.
Developing (2) Uses AI tools but with limited verification. Some signs of over-reliance on unverified outputs.
Emerging (1) AI tools used superficially. No evident verification. Output accepted at face value.

Dimension 2 — Critical Thinking

(VALUE: Critical Thinking)

Level Descriptor
Exemplary (4) Clear articulation of assumptions, limitations, and alternative interpretations. Distinguishes evidence from inference throughout. Identifies what would change the conclusion.
Proficient (3) Generally well-reasoned. Acknowledges limitations. Some evidence of considering alternatives.
Developing (2) Reasoning is present but surface-level. Alternative interpretations not deeply explored.
Emerging (1) Conclusions asserted without visible reasoning. No engagement with alternative interpretations.

Dimension 3 — Ethical Reasoning

(VALUE: Ethical Reasoning)

Level Descriptor
Exemplary (4) Names specific ethical considerations relevant to THIS project (not generic “AI ethics” boilerplate). Addresses them substantively. Identifies who could be harmed and how.
Proficient (3) Ethical considerations identified and addressed. Some specificity. No major ethical blind spots.
Developing (2) Generic ethical concerns mentioned. Limited engagement with specifics of the project.
Emerging (1) Ethics absent, or limited to a perfunctory sentence. Specific harms and protections not considered.

Dimension 4 — Communication

(VALUE: Written Communication, or Oral Communication for the talk)

Level Descriptor
Exemplary (4) Writing/presentation is clear, purposeful, and audience-appropriate. Specific claims backed by specific evidence. The reader/listener knows exactly what you did and why at every point.
Proficient (3) Clear writing/speaking. Mostly concrete. Minor lapses into vagueness.
Developing (2) Understandable but with meaningful vagueness. Reader has to infer what was meant.
Emerging (1) Unclear, disorganized, or generic. Reader cannot tell what the author actually did.

Dimension 5 — Integration & Synthesis

(VALUE: Integrative Learning)

Level Descriptor
Exemplary (4) Connects AI skills to disciplinary knowledge in the author’s field. Synthesizes learning from multiple course modules. Demonstrates transfer to a new context beyond classroom exercises.
Proficient (3) Clear connections between AI skills and field. Uses multiple course concepts.
Developing (2) Applies course concepts but in isolation. Limited synthesis across modules.
Emerging (1) Concepts applied mechanically. No synthesis. Disconnected from the author’s field or goals.

Overall score interpretation

  • Sum of 5 dimensions, out of 20 possible:
    • 17–20: Exemplary overall. This capstone travels well — to graduate school applications, portfolios, interview conversations.
    • 13–16: Proficient. Solid work. The author demonstrates course competencies.
    • 9–12: Developing. Capstone shows partial mastery. Revision recommended.
    • 5–8: Emerging. Significant gaps. Further work needed before the capstone should be used externally.

Benchmarking externally

After you complete your capstone, you can compare your work against nationally-scored student work in the AAC&U VALUE Scoring Collaborative (formerly the VALUE Institute) repository. They publish anonymized samples scored against the same VALUE rubric dimensions your capstone uses.

This is not required for course credit. But if you are applying for competitive internships, graduate programs, or AI-related fellowships, being able to say “my capstone scored at the Capstone level (4) on the AAC&U Integrative Learning rubric” is a concrete credential that employers and admissions committees recognize.

What consistently fails

Patterns I’ve seen cause capstone failure:

  1. Starting without planning worksheet. Time saved up front is lost several times over in week 3 rewrites.
  2. No real user / audience. If the end user is “the instructor,” the capstone won’t survive outside the course. If it’s a specific real person, it will.
  3. Overscoping. Better to do 3 things excellently than 8 things superficially.
  4. Skipping the failure-modes section. Capstones without honest failure-modes signal unreflective work, which is the single biggest ethical red flag in AI.
  5. Avoiding peer review. The students who use peer review feedback finish stronger. The students who ignore it finish weaker. This is extremely consistent.
  6. Working in isolation the whole time. Good capstone work happens in conversation, not alone in a library at 2 AM.

Final quiz

1. The single most important section of a capstone talk is:

  1. The hook
  2. The failure modes / what it gets wrong
  3. The conclusion
  4. The slide design

2. The AAC&U VALUE rubrics are accepted by:

  1. Only West Coast universities
  2. No accrediting bodies
  3. All U.S. regional accrediting bodies
  4. Only private universities

3. The capstone rubric’s Integration & Synthesis dimension most closely maps to which VALUE rubric?

  1. Critical Thinking
  2. Written Communication
  3. Integrative Learning
  4. Global Learning

4. In the criterion-focused peer review protocol, scoring happens:

  1. After writing prose comments
  2. Before writing prose comments
  3. Instead of prose comments
  4. Only for struggling students

5. Capstone Option A (Integration Plan) is best for students who:

  1. Are heading to graduate school
  2. Already have active projects or jobs where AI integration could apply
  3. Want to practice teaching
  4. Prefer theoretical work

6. A capstone scoring 18/20 is at what level?

  1. Emerging
  2. Developing
  3. Proficient
  4. Exemplary

7. The week-2 question most likely to prevent capstone failure is:

  1. “What is my final grade going to be?”
  2. “What is most likely to make me want to rewrite this from scratch, and how do I prevent that?”
  3. “Can I change options at this point?”
  4. “Is my topic interesting enough?”

8. The rubric dimension “Ethical Reasoning” scores as Exemplary when the author:

  1. Includes a generic AI ethics paragraph
  2. Names specific ethical considerations relevant to THIS project and addresses them substantively
  3. Cites three major AI ethics papers
  4. Avoids any ethically fraught use of AI

(Answers: 1-b, 2-c, 3-c, 4-b, 5-b, 6-d, 7-b, 8-b)

After the capstone: what happens next

Your capstone deliverable will be one of the three artifacts in your portfolio from Module 12. Before this course ends, make sure you have done all five of the following:

  1. Posted the capstone deliverable to your portfolio URL. If it’s sensitive (includes student data, proprietary information), post a summary and link to a private-access version.
  2. Written a 100-word summary of the capstone in plain language. This is what goes on your resume.
  3. Logged the rubric score (self-assessed or peer-assessed) in a place you can retrieve in 2 years.
  4. Sent the link to one non-classmate — a professor, a mentor, a family member, a potential employer contact. Work that nobody outside the course sees is half-work.
  5. Scheduled one 30-minute “what I learned” reflection two weeks after course ends. Walk through the capstone with fresh eyes. Note what would be different if you did it again.

Module 14 Summary

Key Takeaways
  1. Choose the option that matches YOUR context — working professionals → Integration Plan; builders → Custom AI Solution; educators → Workshop; researchers → Mini-Research
  2. Scope narrowly — the best capstones solve one specific problem well
  3. Test with real users — even 5 users over 1 week produces meaningful data
  4. Iterate based on feedback — show what you changed and why
  5. Reflect honestly — what surprised you? What failed? What would you do differently?
  6. The capstone is your portfolio piece — it proves to employers and yourself that you can use AI effectively, ethically, and strategically
💭 THINK ABOUT IT

You started this course with a healthy suspicion of AI hype and probably some anxiety about whether AI would erode your field or your career. You finish it with skills that most of the workforce does not have, a portfolio that documents them, and a weekly routine that will keep you current. The 28% salary premium is real. The 78 million net new jobs are real. The opportunity and the risk are both concentrated in the same population: people who can tell hype from substance and act on the difference. You are now one of them. The question is no longer whether you understand AI. It’s what you’ll build with that understanding.

Previous: Module 13 — The Future of AI | Home →