Intunio is a design and development studio in Gothenburg, building AI-ready websites: static, client-owned, with design tokens, shadcn-based components, and Markdown as content. A format that both your team and AI agents can read directly, without plugin or proprietary backend.

Our starting point is many years of experience with token-based design systems and web development, reframed for a world where the code is no longer necessarily written by a human and where content should be versioned, reviewed, and published in the same rhythm as code. Content, design, and code live in the same repository and belong together from the first commit.

When does your team need an AI-ready website?

For teams with an existing CMS-based site and no friction, keeping it alive is enough. Three situations where an AI-ready website becomes relevant:

When agents need to read and update the content

Claude Code, Cursor, and Codex can act directly on Markdown files in a repo in a way they can't against a proprietary CMS database. That opens up AI-driven content management, automatic translation, and content-by-prompt.

When you want to own the architecture without agency lock-in

Your existing site sits in a CMS or hosting stack where you depend on a specific vendor. Changing agency means tearing the product down. You want to own the code, change partners, and keep going.

When content needs to be versioned and reviewed like code

The marketing team wants preview, atomic releases (multiple changes published together), review flows, and history without learning a new CMS UI for every update. Content as Markdown in Git gives the same rhythm as your development process.

In all three cases the point is the same: when everything sits in Git in a readable format, control moves from the vendor back to you.

What we do

• Design tokens as CSS custom properties: colour, typography, spacing, and motion expressed so that a future brand refresh changes in one place, not per component

• Static site with a modern stack: Astro as default (zero JavaScript by default, islands for interactive parts), Next.js or Vue where it fits the product better

• Shadcn-based components committed: buttons, form fields, cards, navigation — in your repo, with your tokens, no npm package dependency

• Content blocks and page templates: a set of reusable blocks (hero, feature grid, quote, FAQ, etc.) that the marketing team combines without developer involvement

• Markdown/MDX with front matter: content as files in Git, schema validation on front matter, schemas matching the page type

• Editorial preview and atomic releases: changes in feature branches, preview deploys per branch, releases that ship when everything is ready together

• Accessibility as default: WCAG 2.2 AA built in from the start, semantic structure, contrasts in the token system

• Edge hosting: Cloudflare Pages, Vercel, or equivalent global CDN. No application server to maintain

Common audiences are startups, platform companies, B2B tech companies, and communication-heavy organisations that don't want to be locked into a CMS stack.

Not just a new stack — a different way of thinking about websites

At Intunio we see AI-ready websites as a natural extension of token work and Git-driven development we have done for years.

What really changes is where the source of truth lives. In a classic CMS stack, content sits in a proprietary database, the design system in a separate package, and hosting behind an agency portal. In an AI-ready website, everything sits in the same Git repo: design tokens, components, content, page templates, and deploy configuration. Whoever has repo access has the whole product — agent or human.

That's also why the client owns the code from the first commit. No proprietary dependencies, no specific agency that has to be involved for maintenance. When the agency is done, your team takes over, or moves to the next agency without tearing the product down.

Is this the right service for us?

This service fits when you want a website that is client-owned, AI-readable, and that your marketing team can manage via Git or a lightweight Git-based editor. It fits particularly well for platform companies, B2B tech, and startups that value control and speed over traditional CMS functionality.

For clients who need editor-driven publishing at scale, translation of heavy content volumes, or integration with existing enterprise stacks, our classic web development service with headless CMS (Payload, Sanity, Strapi, Contentful) is often better suited. CMS flows with roles, reviews, and translation features are still stronger in a traditional stack.

The two services are not mutually exclusive. Many clients start AI-ready in a first phase and add a heavier CMS as editorial needs grow. And the design system is the same in both, just packaged differently.

How we work in an AI-ready web project

Four phases:

Design system and tokens

We set design tokens (colour, typography, spacing, motion) as CSS custom properties and document them as part of the repo. Tokens are the source of truth for all visual consistency. The foundation set of shadcn-based components is also built here, customised to your tokens.

Content blocks and page templates

We design and build the set of content blocks the marketing team will use (hero, quote, feature grid, comparison, FAQ, etc.) and the page templates that combine them. The content model is defined through front matter schemas in MDX files.

Implementation and content flow

Components are built as Astro/Next components against the design system. Editorial preview, atomic releases, and template-driven navigation are set up. CI/CD and edge deploy are configured (client or us, by agreement).

Validation, accessibility, and launch

We run a WCAG 2.2 AA review, performance tests against Core Web Vitals, SEO redirects from the old structure. We do a handover where your team can publish without further help from us.

For clients who want to continue after launch, we move into a rolling monthly engagement for new components, content blocks, performance work, and support when the marketing team runs into new needs.

What we offer

AI-ready websites vary in scope depending on the number of page types, content blocks, and content model complexity. Four common engagements:

• AI-ready website for a startup or smaller company (80–120 hours, delivered in 3–6 weeks): design tokens, 8–12 content blocks, 3–5 page templates, a foundation set of shadcn components, Markdown-based content, edge deploy. Suitable for a marketing site or product landing page.

• AI-ready website for a platform company or product portfolio (from 320 hours, with a typical platform build landing at 500–800 hours; delivered in 2–4 months): a larger design system, 15–25 content blocks, several page types (product, brand, company, news), editorial preview flows, editor support via Git-based editor. Suitable when the site needs to carry a brand and a product portfolio across several surfaces.

• Conversion: CMS-based website → AI-ready (160–320 hours, delivered in 4–8 weeks): you have an existing site on Sanity, Contentful, WordPress, or similar. We migrate content to Markdown, establish design tokens, rebuild components as shadcn-based, and move hosting to the edge.

• Long-term maintenance (rolling monthly, from 40 hours/month): new content blocks, page templates, design adjustments, performance work, and support for the marketing team. Lighter maintenance than traditional web projects because most content work happens in Git by the client themselves.

For clients who need a design system as the foundation for multiple products, we build it in parallel or link to our AI-ready design system service. The two ways of packaging design system and website follow the same principle: everything sits in Git, everything is agent-readable, everything is owned by the client.

Pricing

995 SEK/hour (discounted rate with a monthly agreement)

We apply a discounted hourly rate for monthly agreements: you pay the month's estimated cost in advance, and get a price you can plan around. No commitment beyond the current month.

AI tools are included in the rate

We use Claude, Cursor, and Codex to write component code, generate content block variants, set up page templates, and iterate on design tokens. The AI-ready architecture is also built so your own AI flow keeps working after launch — the agent has the same access to the repo as you do.

Common starting points

Three typical situations where a conversation with Intunio becomes relevant:

When a new brand identity needs to reach the web quickly

You've reworked the brand platform, visual identity, and tone of voice, and need a site that reflects it. We go from brand assets to live website via design tokens and content blocks, often in a few weeks.

When an existing site has outgrown its CMS stack

You notice that updates take too long, the agency dependency grows, and every feature requires a developer. The AI-ready architecture moves control back to you.

When you want to work AI-driven from the start

You already use Claude Code, Cursor, or similar for development work. A site in Git with Markdown content and shadcn components lets the agent work as freely on the website as on your code.

In all three cases the result is the same: a website your team owns, can evolve without tearing down, and which can be updated as easily by a human editor as by an AI agent.