FlyRank AI Fluency · Intern Guide
Week 2 · Frame your work

Make the Work Speak

Your work does not explain itself. This week you turn what you already have, and what you make here, into case studies that prove something, in your own voice, with AI as your interviewer.

Phase: Foundations Estimated hours: 3
The idea this week
  • The work is the evidence. The case study is the lawyer who stands up and makes the case.
  • A screenshot, on its own, is evidence sitting on a table. A stranger glances at it for four seconds and moves on, because nobody told them what they were looking at.
  • A case study tells them: here was the problem, here is what I did, here is what changed. Three beats. That is what turns a screenshot into proof.
  • AI will happily remove the blank page. It will also write generic sludge if you let it. The skill is using it to pull your real story out, then editing hard.

Why it matters

A portfolio is mostly the framing around the work, and the framing is what makes a stranger trust you or scroll past. Beginners put their work up and assume it is obvious why it is good. It is not. A recruiter scrolling past a project has no idea what was hard about it, what you decided, or what it changed, unless you tell them. AI means you never face a blank page again, but only if you use it to get your decisions and results out, not to borrow a generic story.

Use your real work, including the work you make here. You are not inventing fake projects. The thing you build this week, the thing you ship next month, each becomes a case on your site. Two real cases told well beat eight thin ones.

Brief

  1. Write a voice card: five to seven words for how you want to sound (for example "direct, warm, plain, no buzzwords"), and add it to your Claude Project as a standing instruction.
  2. Be interviewed: for each real piece of work (start with what you are making in this internship), have AI ask you about it one question at a time until the problem, your decisions, and the outcome are clear. Answer honestly and messily.
  3. Draft each piece into a case study with three beats: the problem, what you did (and what you decided), what came of it. Draft your bio and contact/CTA copy the same way. Direct AI to use your words and keep your voice.
  4. Edit hard. Read every line out loud; cut anything you would not say or cannot stand behind.

Deliverable

The framed cases for every piece the sitemap calls for, in one document, with your voice card at the top, plus one before/after: a generic AI line next to your edited version. Post it in your track thread.

Evaluation criteria (pass/revise)

  • A framed case exists for each piece the sitemap calls for, not a bare screenshot or a title.
  • Each case has the three beats and could only describe your project.
  • It sounds like a specific person; the before/after shows the difference from generic AI copy.
  • It speaks to the one audience and points at the one action; it is tight, with no "results-driven" filler.
The move this week

Interview yourself, one project at a time

First, set a voice card so every draft sounds like you. Then, for each project, do not ask AI to "describe my work." Ask it to interview you until your decisions and your result are out of you. Then shape the answers into the three beats.

Copy these in order. Swap each {highlighted} bit for your own before you send it.

Prompt · Set your voice card first
Here is my voice card. Use it for everything you draft for me: direct, warm, plain, specific, no buzzwords. Short sentences. No "passionate," "results-driven," or "dynamic." A few words that are mine: {three_words_that_sound_like_you}. If a line sounds like a generic AI bio, rewrite it the way a real person would say it to a friend. Confirm you have it, then we will start.
Prompt · Interview yourself about one project
I want to turn one real project into a case study. Interview me before you write anything. The project: {one_line_on_what_it_is}. Who needs to be convinced: {your_one_person_from_week_1}. Ask me one question at a time: what problem or goal it started from, what I actually did and decided, what I tried that did not work, what the result was (even small), and what I would do differently. Do not write yet. Start.
Prompt · Shape the answers into the three beats
Using my answers above and my voice card, write this project as a short case in three beats: the problem, what I did and decided, and what came of it (keep my honest, even small, result). Use only what I told you, do not inflate anything, and keep my words where you can. End with one line on what I would do differently next time. Then list any spot where you had to guess, so I can fix it.
Prompt · Draft your bio and contact line
Write a two-sentence bio and a one-line contact prompt for my portfolio, in my voice card style. My proof statement: {paste_your_proof_statement}. My one action: {your_one_action, e.g. email me to talk}. The bio should say what I do and who for, not list adjectives. Give me three options, plain and human, no buzzwords. I will pick one.
A worked example

Iris turns her museum signage into a case

Iris from week 1 has no "official" design jobs, but she has the museum signage she redrew. Watch the interview turn it into proof.

AI assistant
AI: What problem did the signage project start from?
Iris: Visitors kept asking staff where the bathrooms and the exit were. The old map was decorative and useless.
AI: What did you actually do, and what did you decide?
Iris: I redrew the map around the three questions people actually asked, dropped the fancy illustration, and taped paper versions up for a week to test before printing.
AI: What came of it?
Iris: The "where is X" questions to staff dropped a lot. I do not have exact numbers, but the front desk noticed.
AI: Here is the case, in your voice:
The problem. Museum visitors constantly stopped staff to ask where the bathrooms and exit were. The map on the wall was pretty and unhelpful.
What I did. I rebuilt the map around the three questions people actually asked, cut the decoration, and tested paper versions for a week before anything was printed.
What came of it. Staff noticed far fewer "where is X" interruptions. Next time I would log the before and after properly so I could put a number on it.

That is a real case, and it could only be Iris's. Notice the honest ending: she names what she would do better. That candor is what makes a stranger trust her.

Dorian ran the same interview on his shop's Instagram. His three beats: the problem (weekend foot traffic was flat), what he did (posted the same product three ways for a month and tracked which post brought people in), and what came of it (the "behind the counter" posts pulled the most visitors, so he leaned in). Different track, identical shape.

For your track. The three beats are the same; the middle changes. A machine-learning intern shows the data and approach and the honest limits. A marketer shows the campaign and the metric that moved. A developer shows what was missing, what they shipped, and links the repo or the live thing.
Examples

Screenshot, then case

What weak looks likeWhat a case looks like
A grid of project images with one-word titles, no context.Each image opens to a short story: the problem, what you did, the result.
"I designed a new app screen.""Users abandoned checkout at step three. I cut it to two steps and removed the surprise fee. Drop-off fell in testing."
"Passionate, results-driven creative.""I make confusing things simple. Here are three times I did it."

Watch out for

  • Letting AI write it. A fully machine-written case is the trap. Interview, then edit. The point is that a real person is behind it.
  • Skipping the result. A case with no outcome, even an honest small one, is a brochure. Keep one real wrinkle in.
  • Buzzwords. "Leveraged," "spearheaded," "passionate." If the AI reaches for them, point it back at your voice card.

What good looks like

  • Each case could only describe your project, not anyone's.
  • The writing sounds like you talking to a friend.
  • A stranger could read one case and say what you are good at.
Words you'll hear this week
  • Case study. One piece of work, framed so a stranger gets why it matters.
  • The three beats. The shape of every case: the problem, what you did, what came of it.
  • Voice card. Five to seven words for how you want to sound, given to the AI as a standing instruction.
All of these are in the Glossary with a plain-language definition.

Linked resources

Nielsen Norman Group

Creating a Case Study ↗

The anatomy of a real case study: the problem, your role, what you did, and the result. Maps straight onto the three beats in this brief.

FreeVideo
 Julian Shapiro

Writing Well Handbook ↗

A 300-hour guide to clear, human writing. Use it to turn messy notes into copy that sounds like you, not a generic AI bio.

FreeGuide