Case 01 — my site
Doneiukhvidov.dev — a site that is its own proof.
A portfolio site built on the very model it sells.
The brief
A portfolio site is the tool that brings me small-business clients in Israel. It has to hold a position — one person, no agency — and show my level before anyone writes. So I built it on the very model I sell.
- 013 locales, HE — RTL
- 02Lighthouse — target 90+
- 03Mobile-first, no AI-generic look
- 04One developer, an AI-amplified stack
Key decisions
Decision 01
Anti-agency as a stance, not as a design
╴ A
Brutalism, a dark techy look — so the contrast lives in the design too.
╴ B
Warm editorial, with the contrast living only in the copy.
I chose path B. To a small-business owner in Israel brutalism reads as cold and expensive. A warm look keeps trust, while the anti-agency stance stays sharp — without shouting in the design.
Decision 02
Services by niche, not by technology
╴ A
4 cards — Websites / Automations / CRM / AI, like any freelancer.
╴ B
Services laid out by business type — barbershop, cafe, carpentry, garage.
I chose path B. A barbershop owner is not looking for web development — he is looking for someone who gets that bookings fall through. A niche instead of a technology lets the client recognize himself at once.
Decision 03
RTL from day one, not a patch at the end
╴ A
Build in English, then add Hebrew support at the end.
╴ B
Design Hebrew first, with logical properties from the first commit.
I chose path B. Hebrew here is the primary language, not supported. RTL designed from the start does not break on numbers, arrows and grids — and does not need a whole separate round of fixes.
Decision 04
Demos instead of descriptions
╴ A
Screenshots and copy about our solution.
╴ B
Working demos embedded in the page — you drag, click, change.
I chose path B. You do not take my word for it — you drag a CRM card right here and now. A live demo proves the level instead of promising it.
Decision 05
Price in chat, not in a package table
╴ A
3 fixed price packages on the page.
╴ B
A lower price anchor plus a direct question on WhatsApp.
I chose path B. In Israel a small-business price is closed in writing. A fixed package is either too expensive for one client or a loss on another — a visible anchor and a conversation are fairer than both.
Decision 06
Performance as a promise, not a record
╴ A
Show a screenshot of one perfect Lighthouse number once.
╴ B
Set the numbers as a target re-measured before every handover.
I chose path B. A single number in a screenshot says nothing about your site. A measured target says what I commit to — including the honest fact that Mobile is still below Desktop.
Metrics
Mobile Performance is still below Desktop — that is in progress, I am not hiding it.
Process
- 01
Concept locked
Anti-agency plus services by niche — the two decisions everything else rests on.
- 02
The home page
Manifesto, niches, live demos — a navigator that pulls the visitor inside.
- 03
Service pages
Websites and CRM built as a chain — Claude and a team of agents, one person leading.
- 04
The work section
The case index and the first case — the page you are reading now.
Stack
Framework
Styling
i18n
Interaction
Language
Not WordPress. Next.js is faster, safer, and hosting costs you 0–10 ₪ a month instead of 50–100. A stack chosen for your wallet, not for mine.
This case is about a website
Want one for your business — Websites and landing pages