How I built Plinth, the studio site itself, in 3 weeks for under $500 in tools.

The setup

I started Plinth in early 2026 with a few constraints:

1. Solo studio. No co-founder, no employees. I'd be designer, developer, copywriter, and salesperson, at least for the first few clients.
2. No agency overhead. No subscription bloat. Every tool had to earn its place.
3. Modern stack. I'm building for SaaS startups; my own site has to be on the stack I'd put their sites on.
4. Fast. I wanted a real site live in under a month so I could start outbound.

What follows is the real cost breakdown, the real timeline, and the choices I'd make again vs the ones I'd change.

The full stack

The tools that built Plinth, with real costs.

Tool	What it does	Cost	Free alternative
Next.js 16	Framework	$0 (open source, MIT)	n/a, this IS the stack
Vercel	Hosting + deployment	$0 (Hobby tier)	Cloudflare Pages, Netlify
Tailwind CSS	Styling	$0	Vanilla CSS
Framer Motion	Animation library	$0	CSS animations
shadcn/ui	Component primitives	$0	Radix UI alone
Lucide React	Icons	$0	Heroicons, Phosphor
Cal.com	Calendar booking	$0 (free tier)	Calendly free tier
Cursor	AI-assisted code editor	$20/mo	VS Code + Copilot
Figma	Design + wireframes	$0 (free tier)	Penpot
Cloudflare	DNS + domain	$10/yr (domain)	n/a
Resend	Transactional email	$0 (free tier)	Postmark trial
Plausible	Analytics	$0 (self-hosted) or $9/mo	Umami self-hosted
Domain (plinthstudio.dev)	Yes	$13/yr	n/a

Total ongoing tool cost: $33/month ($20 Cursor + $9 Plausible + ~$2/mo amortized domain). Total one-time cost: $23 (domain + transfer).

Three weeks of work. Under $500 cumulative spend including some experimentation with tools I didn't end up using. If I had to do it again completely fresh, it'd be even cheaper, I'd drop the $20 Cursor and use VS Code with Copilot at $10/mo.

Week 1: Positioning and design

The single most important week. Almost no code written.

Day 1–2: Positioning work. I sat down and wrote answers to April Dunford's five questions: what's the competitive alternative (Fiverr, freelancers, mid-tier agencies), what's unique about Plinth (transparent pricing, senior-led, ships fast), what's the value (premium-quality startup sites without enterprise overhead), who cares most (funded startups at seed/Series A), what's the category frame (boutique design studio for startups). One page of notes. Half a day to write, half a day to argue with myself about it.

Day 3: Reference collection. I pulled together 30 sites I admired and 10 I actively didn't. Looked at typography, layout, color, motion, copy tone. Found patterns. The references that ended up mattering most: Stripe (italics as emphasis), Linear (product in the hero), Anthropic (cream warmth instead of pure white), Refokus (transparent pricing), Designjoy (pricing in the nav).

Day 4–5: Wireframes in Figma. Low-fi greybox wireframes of the homepage and the pricing page. No colors, no fonts, no images. Just structure. The discipline of getting structure right before pixel-pushing is the single biggest accelerant I know.

Day 6: Visual direction. I picked the type stack (Cormorant Garamond for display headlines, DM Sans for body, system mono for code), the color palette (warm off-white canvas #F0EDE7, deep black-brown #1A1814, gold accent #D4A96A), and the design system primitives. One day. Locked.

Day 7: Hi-fi homepage mock. Full Figma comp of the homepage. The thing I'd build in code week 2.

Week 2: Build

The code week. Mostly fast because of week 1 discipline.

Day 8–9: Next.js scaffold + design system. Set up Next.js 16, Tailwind config matching the Figma design system tokens, Lucide icons, Framer Motion. Got the page chrome (nav, footer) and core components (button, card, callout) shipping. Used shadcn/ui as the component primitive layer, it's not a UI library, it's a copy-paste system that gives you a starting point you actually own.

Day 10–12: Homepage build. From Figma comp to working homepage. Hero animation, scrolling sections, work cards, pricing teaser, FAQ component. Where I lost the most time: the hero animation (Framer Motion variants for sequenced fades). Where I saved the most time: shadcn/ui meant I didn't have to design every button state from scratch.

Day 13: Inner pages. Pricing, About, Cost (the calculator-style page). Component-based meant pages 2 and 3 were 5x faster than page 1.

Day 14: Mobile QA + responsiveness. Real device testing on iPhone, Android, iPad. Fixed about a dozen layout issues. Mobile is 79–82% of SaaS traffic, I was not shipping a desktop-first site.

Week 3: Polish, content, and launch

The week I underestimated.

Day 15–16: Copy pass. Wrote every word on the site. The hero. The pricing tiers. The FAQ. The footer microcopy. This took longer than I expected, about 14 hours total. Lesson: even if you're a strong writer, count copy time separately from design time.

Day 17: SEO + meta. Generated OG images (used a simple Next.js OG image template), wrote meta descriptions, added schema markup (Organization, WebSite, FAQPage where applicable), built the sitemap. Lighthouse audit. Got LCP under 1.2s on mobile.

Day 18: Integrations. Cal.com booking embed. Resend for the contact form. Plausible analytics. Each one took roughly an hour. Modern tooling is genuinely good now.

Day 19: Cross-browser QA. Chrome, Safari, Firefox, Edge. iOS Safari and Android Chrome on real devices. Found two Safari-specific layout bugs (the eternal flex-gap issue), fixed them.

Day 20: Final review. Read every page out loud. Caught three typos and one inconsistent capitalization. Reordered two sections that read better the other way.

Day 21: Launch. Pointed DNS to Vercel, deployed production build, sent a quiet LinkedIn post. Site live.

The week-3 lesson

Week 3 is where most solo projects die. The design + build feels mostly done by end of week 2, and the polish work feels small. It isn't. Copy, SEO, integrations, QA, and launch each eat a day. Plan for a full week 3 even when it doesn't feel needed.

The decisions I'm glad I made

Five choices that paid off.

1. Next.js + Vercel over Webflow. I considered Webflow for about 30 minutes. The deciding factor: as a developer, I can iterate on Next.js faster than I can fight Webflow's editor. Also: zero vendor lock-in, complete performance control, the option to add real backend features later (booking flow, payments, AI tools).

2. shadcn/ui as the primitive layer. Not a UI library, a copy-paste component system. I own every line of the components in my repo. No version lock-in, no breaking changes from a third party, full styling control. The right primitive for studio-quality work.

3. Cormorant Garamond + DM Sans pairing. Most premium SaaS sites are using Inter, Söhne, or a custom typeface. Picking a slightly warmer pairing (Cormorant has actual personality, DM Sans is geometric but friendly) gave Plinth a visual signature in a category dominated by Inter.

4. Italicized phrases in H1s and H2s. Stripe taught me this. One simple emphasis system, italics in specific phrases, creates visual rhythm across long pages without resorting to bold or color.

5. Transparent pricing in the nav. Pricing is one click away from anywhere on the site. Filtered out tire-kickers within a week of launch. Inbound that does come through is qualified.

The decisions I'd change

Three things I'd do differently.

1. I'd write the copy in week 1, not week 3. I designed the site first, then wrote copy to fit. The copy ended up reshaping the design. If I'd written the copy first, I'd have done one fewer design pass.

2. I'd build a CMS sooner. I'm hardcoding blog posts as MDX files right now. That's fine for the first 10 posts. For the next 50, I should have set up Sanity or Payload in week 2. Migration is easy in theory but I keep delaying it.

3. I'd ship a smaller v1. I built homepage + 4 inner pages for launch. I could have launched with homepage + pricing in week 2 and added the rest in week 4 while running outbound. Speed-to-first-client matters more than completeness.

The honest cost ratio

Here's what 3 weeks of solo work would cost a client at Plinth pricing:

• Strategy + positioning (1 day): ~$1,500
• Wireframes + design (3–4 days): ~$6,000
• Build (5–7 days): ~$8,000
• Copy (2 days): ~$2,000
• SEO + meta + integrations (1 day): ~$1,500
• QA + launch (1 day): ~$1,500

Total equivalent value: roughly $20,500. Which lines up with Plinth's Authority tier at $22,000. That's a useful sanity check, what I'd charge a client for this scope is what the project actually cost me in time at fair hourly rates.

The reason I could do it for $500 in tools: I'm not paying myself for the time. Founders building their own site are typically paying themselves nothing, then mistaking that for "the site cost nothing." The opportunity cost is real, three weeks I didn't spend on outbound, on building Plinth's pipeline, on closing a client.

For most founders, the math favors hiring a studio: $5K–$25K spent, but you keep your 3 weeks for the work only you can do. I built my own because I'm a designer and developer running a design studio, not building it would have been weird. For everyone else, that calculus probably points the other way.

The tools you actually need

If I were giving a founder my shortest possible "build it yourself" stack today:

1. Next.js + Vercel, the foundation. Free.
2. Tailwind + shadcn/ui, styling and components. Free.
3. Cursor ($20/mo) or VS Code + Copilot ($10/mo), AI-assisted editing.
4. Figma (free tier), design.
5. Cal.com (free), booking.
6. Resend (free tier), email.
7. Plausible ($9/mo), analytics that don't require GDPR cookie banners.
8. A domain (~$15/yr).

Total: roughly $30/month + a domain. For 10x what most agencies will charge you in setup fees, you can run the same stack used by Vercel, Linear, Notion, and most of the YC W26 batch.