FlyRank AI Fluency · Intern Guide
Week 4 · Pick the stack

Options, Not Decrees

This week you choose how to build and host your site, by making AI lay out real options with trade-offs instead of one confident answer. Then you create the empty project, live on a URL. This is a skill you will use far beyond websites: deciding with AI while staying the one in charge.

Phase: Build Estimated hours: 4
The idea this week
  • People think this is where they finally need to be an engineer. They do not. They need to make one good decision.
  • Do not bring a bulldozer to plant a flower. The goal is not the impressive tool, it is to finish and to show your work well. The best stack is the smallest one that does the job and that you can maintain.
  • The move is not "AI, build my site." It is "AI, give me three options with the real trade-offs," then you choose with your eyes open.
  • The week ends in a small, strange, powerful milestone: empty but live. A blank page already on a URL.

Why it matters

Choosing how to build is itself an AI-fluency skill, maybe the most useful one. The win is not getting AI to pick for you; it is giving it your real constraints, making it lay out options with trade-offs, and deciding for yourself with eyes open. Treating AI as an adviser you interrogate, not an oracle you obey, is core to being fluent with it. And "empty but live" is a real milestone: the hardest step is going from nothing to something on a URL. Do that now and next week you are filling a thing that already exists.

Bring one extra thing to the conversation: how your work needs to be shown. A designer needs clean image galleries. A machine-learning intern needs to embed a live demo or a notebook. A writer needs readable long-form. That single detail changes the right answer, so say it out loud.

Brief

  1. Give AI your four constraints: free only, your honest skill level, what your portfolio needs to do (paste your sitemap and content map), and how your work must be displayed (image galleries, an embedded demo, a code repo, long-form reading). State whether anything has to be dynamic yet.
  2. Make it produce three stack options, simplest to most powerful, each with: how you would build, where you would host for free, whether it needs a backend, and the real trade-off.
  3. Pressure-test the front-runner: what breaks if I pick the simplest, what do I maintain if I pick the most powerful, can I finish in time, does it show my work the way it needs to be shown?
  4. Decide, and write a short rationale in your own words: the stack you chose, the two you did not, and why, including "can I maintain this."
  5. Create the empty project for your chosen stack: start the no-code project, or create a repo and connect a free host (GitHub Pages, Netlify, Cloudflare Pages, Vercel) so a near-blank page is reachable on a URL. Confirm it is really live by opening it on your phone.
  6. Drop your identity kit, case studies, and content map into your Claude Project so the build week has everything in one place.

Deliverable

The written rationale (chosen stack, the alternatives considered, and why), plus the live URL of the empty or near-blank project (even if it just says your name) and a screenshot. Post it in your track thread.

Evaluation criteria (pass/revise)

  • Three genuine options with trade-offs were considered, not one answer obeyed.
  • The chosen stack is free, matched to real needs, and displays your kind of work properly.
  • The rationale is in your own words and includes "can I maintain this."
  • The backend question is answered honestly (most will correctly choose "not yet").
  • A real, reachable URL exists, opened on a second device to prove it, and it matches the chosen stack.
  • The Claude Project has the identity kit, case studies, and content map loaded for next week.
The move this week

Ask for three roads, then walk one

There are three broad roads. No-code (drag and drop, genuinely right for many portfolios). Plain code with AI on a free host (the common default: truly yours, free forever, and you learn how a site works). A framework (real, and usually overkill). Make AI lay them out for your situation, then choose.

Run these in order: get options, understand the choice, then set it up. Swap each {highlighted} bit first.

Prompt · Get your three stack options
Help me choose a stack for my portfolio. Do not pick for me, give me options. My constraints: free tools only, I am a beginner who is learning, I must be able to maintain and explain it. What it needs to do: {paste_your_sitemap_and_content_map}. How my work must be shown: {image_galleries / a_live_demo / a_code_repo / long-form_reading}. Does anything need to be dynamic at launch (a contact form), yes or no? Give me three options, simplest to most powerful, each with: how I would build it, where I would host it free, whether it needs a backend, how well it shows my kind of work, and the real trade-off. End with the one you would pick for me and why. Then I will decide.
Prompt · Explain the choice in plain words
I am leaning toward this option: {the_option_you_like}. Before I commit, explain it to me like I will have to defend the choice to someone: what each part does, what I will have to learn, what could trip up a beginner, and what it would cost me to switch later if I outgrow it. Then ask me two questions to make sure it really fits my work.
Prompt · Set it up, step by step, to "empty but live"
I have chosen this stack: {your_chosen_stack}. Walk me through setting it up from zero, one step at a time, waiting for me to confirm each step works before the next. The goal for today is small and specific: a blank page that shows just my name, live on a free URL. Tell me exactly what to click or paste, and how to check each step worked. Assume I have never done this before.
A worked example

Iris picks the road that shows her screens well

AI assistant
Iris: I almost picked a heavy framework because it looked professional.
AI: For a three-page portfolio you would spend your two weeks fighting build errors instead of showing your work. What does your work need most?
Iris: Clean, big screenshots of my redesigns, and maybe one clickable prototype.
AI: Then two good roads: a no-code builder like Framer (great image galleries, fast to publish), or plain HTML and CSS that we write together on a free host. The second is more yours and teaches you more.
Iris: Let us do plain code with you. I want to understand it.
AI: Good. Here are the steps to create an empty repo, connect a free host, and get a blank page with just your name live on a URL. We fill it next week.

Iris did not pick the most advanced tool. She picked the smallest one that shows her work well and that she can finish. Then she got "empty but live," which closes the scariest gap of all: nothing to something on the internet.

Dorian's work is words and numbers, not screens, so the three-options prompt steered him somewhere else: plain code on a free host, because clean, fast, readable long-form was what his case studies needed. Same prompt, the road that fit his proof.

For your track. Let the work pick the road. Machine learning: make sure the host lets you embed a demo or notebook and link a repo. Marketing: favor clean, fast, readable long-form. Dev: code on a free host doubles as proof you can ship, and a tidy repo is part of the portfolio.
The three roads, at a glance

All free, all real

RoadWhat it isGood when
No-codeDrag and drop in a builder (Carrd, Framer, Canva).You want to publish fast and your work is mostly visual.
Code with AIPlain HTML and CSS, written with the AI, on a free host (GitHub Pages, Netlify, Cloudflare, Vercel).You want it truly yours, free forever, and to learn how a site works. The default.
FrameworkA full developer setup.Almost never, for a portfolio. Only if the site genuinely needs it.

Watch out for

  • Choosing the impressive tool. The "do it properly" instinct is the one most likely to stall you. Finish beats fancy.
  • One confident answer. If the AI gives a single decree, send it back: three options with trade-offs.
  • Skipping "how is it shown." A stack that cannot display your evidence well is the portfolio-specific mistake.

What good looks like

  • You can say in one sentence why you picked your stack.
  • A blank page with your name is live on a URL, confirmed on your phone.
  • You feel a little proud of "empty but live." You should. The hardest gap is closed.
Words you'll hear this week
  • Stack. The set of tools you use to build and run your site.
  • No-code. Building with a visual drag-and-drop tool, no code written.
  • Host / hosting. The free service that puts your site online at an address.
  • Repo (repository) / GitHub. The folder that holds your project and remembers every version.
  • Deploy / publish. Pushing your latest version live.
  • Backend. The part that remembers or does something a plain page cannot. Most portfolios need exactly one, later.
Every term is in the Glossary.

Linked resources

MDN

Getting started with the web ↗

What a website actually is and the few pieces it takes. Read this before you choose anything, so the options make sense.

FreeTutorial
 GitHub

GitHub Pages ↗

Free hosting straight from a repo. The simplest "plain code, live on a URL" path, and your repo doubles as proof you can ship.

FreeHost
 Netlify

Netlify Docs ↗

Free hosting with drag-and-drop deploys, built-in forms, and easy custom domains. A friendly default for this build.

FreeHost
 Cloudflare

Cloudflare Pages ↗

Free static hosting on a fast global network, with unlimited bandwidth. A solid alternative if your traffic grows.

FreeHost
 Carrd

Carrd ↗

If code is not your path: a genuinely good no-code builder for a sharp one-page site. The free tier covers most portfolios.

No-codeFree tier