A portfolio website used to take a weekend of templates and CSS wrangling. With AI it takes an afternoon, and you barely touch code. This guide shows exactly how to make a portfolio website with AI in four steps: ChatGPT interviews you and writes the content, tellnova builds the site, and a deploy button puts it on the internet. No framework choice, no hosting setup, no credit card beyond the AI subscription you probably already pay for.
The workflow works even if you have never shipped a website. You bring your story and your work; the AI brings the HTML, CSS, and deployment plumbing.
What you need
- A ChatGPT account — the free tier is enough for the content step
- tellnova, the desktop app that turns prompts into a running website
- Any AI subscription tellnova supports: ChatGPT (Codex), Cursor, or an OpenRouter key that unlocks GLM, Kimi, MiniMax, and hundreds of other models
- A free Vercel account for hosting — created during the deploy step
That last requirement is the point. Everything else in the stack is either free or something you already have.
Step 1: Let ChatGPT interview you
The mistake most people make is asking an AI to "make me a portfolio" cold. You get generic filler, because the model knows nothing about you. Flip the direction: give ChatGPT your raw material and make it interview you before it writes a word.
Open ChatGPT and paste this:
I want to build a personal portfolio website. Here is my introduction:
<paste two or three sentences about who you are>
Here is my work so far:
<paste links, repos, screenshots, or descriptions of projects>
Before you write anything, interview me. Ask me one question at a time
until you can produce: a one-line headline, a short bio, a write-up for
each project covering the problem, what I built, and the result, and a
skills list. Then output everything as structured markdown I can paste
into a website builder.
Expect ten to fifteen questions. Why did you pick that stack? What broke, and how did you fix it? What would you do differently now? Answer honestly — those answers become the project stories that make a portfolio read like a human wrote it, because one did.
Copy the final markdown. That is your site content, done.
No work to show yet?
If you have not built anything, this step takes a weekend instead of a lifetime: build something small with the same AI stack. Pick a problem from your own life — a study tracker, a recipe archive, a tool for your class or team — and have tellnova build version one in an evening. Ship it, use it, fix it. That single finished project is enough material for the interview above, and the story of how you built it with AI is itself worth a paragraph on the site. The example portfolio this guide ships with is open source — read its content file to see what "enough material" looks like in practice.
Step 2: Set up tellnova
Download tellnova from tellnova.com and sign in. Then connect the AI subscription you already pay for in Settings:
- ChatGPT subscription — sign in once, use it as the build engine
- Cursor — same idea, if that is your subscription
- OpenRouter — bring your own key and pick from GLM, Kimi, MiniMax, Claude, and hundreds more
tellnova works with all of them; the steps below do not change. If you use the ChatGPT route, the walkthrough of using a ChatGPT subscription in tellnova covers the sign-in in detail.
Step 3: Generate the site
Start a new task in tellnova, paste the markdown from ChatGPT, and add one instruction:
Build a single-page portfolio website from this content. Clean modern
design, mobile-friendly, fast to load. Sections: hero, about, projects,
skills, contact. Serve it as a simple static site.
tellnova creates an isolated git worktree and the agent builds the whole site — files, styles, and content. You watch it work, then review the changes as a diff before anything touches your machine. Ask for changes in plain language: "make the hero darker", "add a dark mode toggle", "make project cards clickable". Each follow-up runs in the same workspace, so the site evolves instead of restarting.
If you would rather compare approaches first, the guide on the best AI coding agents in 2026 covers how different agents handle jobs like this one.
Step 4: Deploy with one click
This is where tellnova quietly removes the step that usually kills beginners: getting the site onto the internet.
When the preview looks right, open the deploy menu on the task and choose Deploy to production. tellnova ships with Vercel's CLI built in; sign in to a free Vercel account once, click, and your portfolio is live on a real URL. Need to test a risky change first? Create preview deployment publishes a draft URL without touching the live site.
Using any other AI agent instead? You will have to describe the deployment yourself — install the Vercel CLI, prompt the agent to configure it, and debug the prompt when it guesses wrong. tellnova just has the button.
Make it yours
A deployed portfolio is a starting point, not a monument. The AI workflow makes iteration nearly free:
- Add each new project as a follow-up task: "add a Projects section entry for X, using this repo"
- Point a custom domain at the Vercel deployment — a settings page, not a project
- Re-run the ChatGPT interview every few months; your story changes
The live portfolio built for this guide is here: https://demo-portfolio-mu-ochre.vercel.app. Fork the open-source repo and make the content yours.
FAQ
Is making a portfolio with AI free?
The site itself costs nothing: tellnova's deploy uses Vercel's free tier, and the content step fits in ChatGPT's free plan. You only pay for the AI subscription that builds the site — and most readers already have one.
Do I need to know how to code?
No. You review a diff and describe changes in plain language. Reading the code the agent writes is how you learn, not a requirement.
Which AI should I use to build the portfolio site?
Whichever subscription you have. ChatGPT writes the content, and the build step works the same on Codex, Cursor, or any OpenRouter model. For a simple static portfolio, a mid-tier model is plenty; spend the effort on the writing instead.
Can I use this workflow for more than a portfolio?
That is the real takeaway. Interview-and-build scales to landing pages, internal tools, and weekend apps — the portfolio is just the first one you ship.
