How to Build a Personal Website That Does Not Look Generic
Most personal sites fail for the same three reasons. Here is what actually works, plus how to build one with an AI coding tool and still end up with something that looks like you.
Everyone tells you to have a personal website. Almost nobody tells you what should be on it, so most people end up with the same page: a headshot, a paragraph starting with "I am a passionate," three project cards, and a contact form nobody fills in.
That page is not bad. It is just interchangeable. Someone reads it, learns nothing they could not learn from your LinkedIn, and closes the tab.
This guide covers what a personal site is actually for, the structure that works, and how to build one with an AI coding tool without the result looking like every other AI-built site.
What a personal website is for
Be honest about the job before you pick a design. There are really only four:
Getting hired. A recruiter or hiring manager lands on it after seeing your name somewhere. They have about thirty seconds and one question: is this person good at the thing I need.
Getting clients. Someone was referred to you and wants to check you are real before they email. They are looking for proof and a way to contact you.
Being findable. You want your own name to return your own page, not a stale profile on a platform you abandoned.
Having a home for your work. Writing, projects, or experiments that would otherwise be scattered across platforms you do not control.
These want different pages. A hiring page front-loads evidence. A client page front-loads outcomes and a booking link. A writing home front-loads the writing. If you try to do all four at once you get the generic page, because the generic page is the average of all four.
Pick one. The others can be links.
The structure that works
Almost every good personal site follows the same shape. It is not a rule, it is just what happens when the page is honest about the reader's thirty seconds.
One line that says what you do. Not a tagline. A sentence a stranger could repeat to someone else. "I build design systems for fintech companies" beats "Designer. Thinker. Coffee enthusiast."
Proof, immediately. The single strongest piece of evidence you have, above the fold if possible. A product people have heard of, a number, a client name, a screenshot of something real. Not a list of skills. Everyone claims skills.
Two or three things in depth, not ten in passing. A page with three projects that each explain the problem, your decision, and the result is worth more than twelve thumbnails. Depth is what a thumbnail grid cannot fake.
A way to reach you that is one click. An email address in plain text beats a contact form. Forms add friction and a lot of people distrust them.
Something that is only yours. A strong opinion, an unusual project, a piece of writing, a detail of how you work. This is the part that makes you memorable rather than qualified. It is also the part every template strips out.
That last one is worth dwelling on. Qualified people are common. Memorable ones are not.
Why AI-built personal sites look generic
If you build your site by prompting an AI coding tool, you will probably get something clean and completely forgettable. This is not the tool failing. It is the tool doing exactly what you asked.
A model without visual direction produces the most statistically likely design: a centered hero, three feature cards, a lot of vertical padding, and a blue or purple accent. That layout appears in its training data a hundred thousand times. Absent other instruction, it is the safest answer.
You get a generic site for the same reason you get generic ideas from a generic prompt. There were no constraints, so it returned the average.
There are three ways out.
Describe the feeling, not the components. "Editorial, lots of whitespace, one accent colour, type does the work" gives more direction than "make it modern and clean." Modern and clean is what it was already going to do.
Point at a reference. Naming a site whose look you admire gives the model an anchor. The catch is that it only knows the site if it is well known, and it will approximate rather than match.
Give it a design system file. Rather than describing style in every prompt, hand the tool a file that carries the decisions: colours and their roles, type scale, spacing, component behaviour. A DESIGN.md does this, and because the file persists, page two looks like page one.
The third option is the only one that survives past the first screen. Prompts are forgotten between sessions. A file is not.
Building it, step by step
Assume you are using an AI coding tool like Claude Code, Codex, or Cursor.
1. Write the content first, in a plain text file. Every section, in final words. This feels like a detour and it is the highest-leverage thing you can do. Design decisions get easy once the words are fixed, and impossible while they are not.
2. Choose a visual direction before you build. Pick a reference or a DESIGN.md whose look matches what you want to say. A research-heavy personal site and a bold creative portfolio need different systems, and choosing after you build means rebuilding.
3. Build one page, not a site. Get the single most important page finished and good. Most personal sites should stay one page. See the section below on when a second page is warranted.
4. Read it as a stranger. Open it on your phone, imagine you have never met yourself, and see whether the first screen answers "what does this person do and are they any good." If not, the problem is the first line, not the design.
Stop building from a blank page. LaunchKit gives your AI coding assistant a complete, working site to start from, so the first screen you touch is the one that makes it yours.
5. Ship it before it is finished. A live site with three sections beats a perfect one that is still local in six months. You will edit it more often once it is real.
When one page is enough, and when it is not
Most personal sites should be one page. All the content in one scroll, no navigation to design, no decisions about hierarchy between pages.
Add a second page only when a specific thing has outgrown the scroll:
- Your writing, once there is enough of it to deserve its own index
- A project that needs a full case study rather than a paragraph
- A CV that people ask for as a standalone link
If you are adding pages because the site feels too short, do not. A short site that says something is better than a padded one that does not. We covered the one-page tradeoff in more depth in the one page website guide.
A quick before and after
Before:
Hi, I'm Alex. I'm a passionate full-stack developer with experience in React, Node, and cloud technologies. I love building things and solving problems.
This tells you nothing. Every developer says it.
After:
I'm Alex. I build the payment and billing side of B2B products. Last year I rebuilt checkout for a company doing $40M a year and cut failed payments by a third.
Same person. The second version is specific, has evidence in it, and gives someone a reason to email. It also implies a niche, which sounds limiting and is in fact the thing that gets you hired.
The design did not change between those two. Most sites that feel generic have a copy problem wearing a design costume.
Frequently asked questions
Do I need a personal website if I have LinkedIn?
If you only want to be findable, no. A profile does that. You need your own site when you want to control the framing, publish things that do not fit a feed, or show work that a profile flattens into bullet points.
What should be on the homepage?
One line about what you do, your strongest proof, two or three things in depth, and a way to contact you. Everything else is optional.
Should I include a blog?
Only if you will write. An empty blog with one post from two years ago is worse than no blog, because it reads as abandoned. Link to writing you publish elsewhere instead.
Is a custom domain worth it?
Yes, and it is the cheapest credibility available. Your own name is best. If it is taken, adding your field usually beats adding numbers.
How do I stop my AI-built site looking like a template?
Give the tool a design system before you ask for screens, and write specific copy. Those two fix almost all of it. Visual polish cannot rescue a page that says nothing.
How long should this take?
A weekend, if the copy is written. Most of the months people spend are spent avoiding the copy.
Start with the sentence
The hardest part of a personal website is not the build. It is deciding what you want a stranger to know about you in thirty seconds, then having the nerve to say only that.
Write the sentence first. Pick a visual direction that matches the tone of it. Then let your AI tool do the part it is genuinely good at, which is turning fixed decisions into working pages.
