FlyRank AI Fluency · Intern Guide
Week 1 · Decide what you're proving

AI n-able Yourself

Solve for your n. This week you decide, on purpose, what your portfolio is going to prove, and to whom. No code yet. Just the one decision everything else is built on.

Phase: Setup Estimated hours: 5
The idea this week
  • Your mentor runs AI Enablement at FlyRank. Enable. AI n-able yourself.
  • In math, n is the variable: it stands for an unknown, a count, an index, whatever the moment needs.
  • Right now, what you can do with AI is your n: undefined, full of potential, waiting to be solved for.
  • This week we define it. We pick the one thing you are going to prove. Everything in the weeks after is just building the proof.

Why it matters

A portfolio with no claim is just decoration, and decoration convinces no one. Your portfolio has exactly one job: to make a specific real person believe you can do a specific thing. The single most useful hour of this track is the one where you decide what this thing proves and to whom. Everything downstream gets easier once that is settled, and impossible until it is.

A useful test. Can a visitor land on your finished site, tell what you do, believe you are good at it, and find the one thing you want them to do, all in under 30 seconds? If your idea makes that hard, it is still too broad. Narrow it.

Brief

  1. Write a one-paragraph proof statement that answers three questions plainly: the one claim (one primary skill, named), the one person (a specific person who could hire or engage you, not "people"), and the one action (the single most important thing you want them to take). Use AI as a thinking partner, not an author: have it interview you with sharp questions until the claim is genuinely narrow and true. Add one honest line on why this needs to exist: what a CV or LinkedIn alone cannot prove that owning this fixes.
  2. Sketch a small sitemap: the few pages it takes to walk your one person from landing, to believing you, to taking the action. Resist adding pages just because other sites have them.
  3. Set up your free toolkit: accounts for Claude, ChatGPT, Gemini, and Perplexity.
  4. Create one Claude Project named for this build, with custom instructions: who you are, your proof statement pasted in, and a request to act as a tutor that explains its reasoning. It follows you for all ten weeks.
  5. Run one real prompt inside it: have it pressure-test your sitemap against your claim and your one action. Save the answer and note at least one thing you will change.

Deliverable

Your one-paragraph proof statement plus the one-line why, a photo of your sitemap sketch, and a screenshot of your configured Claude Project with the pressure-test prompt and output. Post it all in your track thread.

Evaluation criteria (pass/revise)

  • One primary claim is named, not three skills hiding behind "and."
  • The audience is one specific person who could hire you, not "everyone," and a single most-important action is chosen.
  • The statement could only describe you, not any portfolio.
  • The sitemap is small and every page earns its place against the claim and the one action.
  • The Claude Project has genuine custom instructions (proof statement pasted in), not the defaults.
  • The sitemap was pressure-tested and at least one change is noted.
The move this week

Do not ask AI to write your site. Ask it to interview you.

The biggest beginner mistake is "AI, write my about page." You get something smooth, generic, and forgettable, the same sludge on ten thousand other pages. The skill is the opposite: use AI to pull your real answer out of you, then you decide. We call it the narrowing interview.

Copy any prompt below, paste it into your AI, and swap the {highlighted} bits for your own.

Prompt · Start the narrowing interview
I am switching into {your_field} and building a portfolio whose only job is to prove I can do one thing well, so someone will hire or work with me. Interview me to find three things: the ONE claim I am proving, the ONE specific person I am proving it to, and the ONE action I want them to take. Ask me one sharp question at a time, push back when I am vague or trying to prove more than one thing, and after about eight questions propose a one-paragraph proof statement. Start with your first question.
Prompt · Sharpen a statement that is still too broad
Here is my draft proof statement: {paste_your_draft_statement}. It still feels generic. Give me three sharper versions that each name ONE claim, ONE specific person who could hire me, and ONE action. For each, point out the exact word that was too broad and what you replaced it with. Do not invent achievements I did not mention. Then ask me the one question that would make it sharper still.
Prompt · Pressure-test your sitemap
This is my proof statement: {paste_your_proof_statement}. This is my sitemap: {paste_your_sitemap}. Does this sitemap walk my one person from landing, to believing me, to taking my one action, and does it actually prove my claim? Tell me which pages earn their place, which are wasted, and what is missing. Be specific, not encouraging.
A worked example

Iris is switching into UX design. Here is how her interview went.

Iris used to give tours at a museum. She wants a design job but has no "official" experience, so her first instinct is the vague version. Watch how the interview narrows it.

AI assistant
Iris: I want a site that shows I am creative and can do design, so I can get a job.
AI: "Creative" could mean a hundred things. When you say design, what kind of problem do you most want to be hired to solve?
Iris: Honestly, at the museum I kept redrawing the signs and the map because people got lost. I like making confusing things simple.
AI: That is a real and specific claim. Who is the one person who needs to believe it? A recruiter at a big agency, or someone else?
Iris: Probably a hiring manager at a small product team, the kind of place that would take a chance on someone switching in.
AI: And when that person is convinced, what is the single most important thing you want them to do?
Iris: Email me to set up a chat.
AI: Here is your proof statement: "I turn confusing, real-world flows into simple, usable screens. I am proving this to a hiring manager at a small product team, so they will email me to talk."

Notice what happened. Iris did not borrow a generic line. The interview pulled out something only she could say (the museum signs), and turned "I am creative" into a claim a hiring manager actually recognizes. That is the whole move.

A different track, same move. Dorian is switching into marketing from a retail job. His first version was "I do social media and content." The same interview pulled out the real story: he ran the Instagram for the shop he worked at and grew weekend foot traffic. His statement became "I grow a small local brand's organic reach and can show the before and after, for a founder who wants more of the right visitors, so they will book a call." Same eight questions, his field.

Your track will sound different, and that is the point. A machine-learning intern proves "I build models that work on real data and I know their limits." A marketer proves "I can move a metric and show the work behind it." A developer proves "I can take an idea to something running in production." Same exercise, your field.
Examples

Too broad, then sharp

The difference between a statement that gets revised and one that passes is almost always specificity. Use these as a mirror.

TrackToo broad (revise)Sharp (pass)
UI / UX"I am a creative designer who loves clean design.""I turn confusing flows into simple, usable screens, for a hiring manager at a small product team."
Machine learning"I am passionate about AI and data.""I build models that work on messy real data and I am honest about their limits, for a lead who needs a useful prototype, not a demo."
Marketing"I do social media and content.""I grow organic traffic and can show the before and after, for a founder who wants more of the right visitors."
Dev / SEO / Ops"I am a hard-working problem solver.""I take a rough idea to a working tool in production, for a small team that needs to ship, not plan."

Watch out for

  • The "and" trap. "Design and code and write and market." Pick the one thing you most want to be hired for. The rest can be supporting evidence later.
  • Jumping to the look. The design is week three. A gorgeous site that claims nothing fails the whole point.
  • Letting AI decide for you. It interviews, it suggests. You choose. The claim has to be one you can stand behind.

What good looks like

  • You can say your proof statement out loud in one breath.
  • A friend who hears it knows exactly what you do and who it is for.
  • Your sitemap has three or four pages, and you can say why each one exists.
Words you'll hear this week
  • Proof statement. One paragraph that names your one claim, your one person, and your one action. The foundation of the whole build.
  • The one action. The single most important thing you want a visitor to do: email you, hire you, download your CV, book a call, try your thing.
  • Audience. The one specific person you are trying to convince. "A recruiter at a mid-size studio," not "people."
  • Sitemap. A simple map of your site's pages and how a visitor moves from landing to the action.
  • Call to action (CTA). The button or link that invites the visitor to take the one action.
  • Claude Project. A saved AI workspace with your context (your proof statement, later your stack and style) so it remembers your build instead of starting cold each time.
Every term here lives in the Glossary too, with a plain-language definition you can search any time.

Linked resources