Four stages. You know what happens next at every point, and what it costs.
No hourly billing, no change orders appearing halfway through, no handover that leaves you dependent on us.
- 01Scopeweek 1
- 02Designweeks 1 to 2
- 03Buildweeks 2 to 5
- 04Launchweeks 4 to 6, plus 30 days
The weeks shown are a typical website build. An application or an AI system is scheduled in its own written scope, before anything starts.
- 01week 1
A call, then a written scope. We work out what the site or system actually has to do, who it is for, and what would count as it working.
You get a fixed price and a date. Both are on paper before any money moves on websites and web applications; AI systems start with a paid scoping phase. If the scope changes later we requote before doing the work, not after.
What we need from youwhatever exists already — brand materials, content, access to the systems it has to talk to. If none of that exists, say so. Gemnye sent us one paragraph about their company, and that was enough.
- 02weeks 1 to 2
Tokens and components first, then pages. You see the design system before anything is composed from it, because a wrong decision at that level propagates into everything after.
Two rounds of revision are included. A third is a change of scope, quoted before it starts.
What we need from youone decision at the end of each round. Consult whoever you need to, then send us the consolidated answer.
- 03weeks 2 to 5
We build against the design system, not against screenshots. Tests are written as we go, not at the end.
You get a staging URL from the first week of the build. Look at it whenever you want.
What we need from youfinal content, if you are supplying it, by the date the scope sets. If you would rather not write it, we can.
- 04weeks 4 to 6, plus 30 days
Before anything goes live: the full test suite, an accessibility audit at seven widths, and a Lighthouse run. We publish the numbers rather than describing them.
Then handover. Repository, hosting, domain, analytics, every account in your name. A written record of how it works and how to change it.
Thirty days of support afterwards, included. Bugs are ours to fix. New features are a new project.
What we need from youNothing. This stage is ours.

- 01
Scope
week 1A call, then a written scope. We work out what the site or system actually has to do, who it is for, and what would count as it working.
You get a fixed price and a date. Both are on paper before any money moves on websites and web applications; AI systems start with a paid scoping phase. If the scope changes later we requote before doing the work, not after.
What we need from youwhatever exists already — brand materials, content, access to the systems it has to talk to. If none of that exists, say so. Gemnye sent us one paragraph about their company, and that was enough.

- 02
Design
weeks 1 to 2Tokens and components first, then pages. You see the design system before anything is composed from it, because a wrong decision at that level propagates into everything after.
Two rounds of revision are included. A third is a change of scope, quoted before it starts.
What we need from youone decision at the end of each round. Consult whoever you need to, then send us the consolidated answer.

- 03
Build
weeks 2 to 5We build against the design system, not against screenshots. Tests are written as we go, not at the end.
You get a staging URL from the first week of the build. Look at it whenever you want.
What we need from youfinal content, if you are supplying it, by the date the scope sets. If you would rather not write it, we can.

- 04
Launch
weeks 4 to 6, plus 30 daysBefore anything goes live: the full test suite, an accessibility audit at seven widths, and a Lighthouse run. We publish the numbers rather than describing them.
Then handover. Repository, hosting, domain, analytics, every account in your name. A written record of how it works and how to change it.
Thirty days of support afterwards, included. Bugs are ours to fix. New features are a new project.
What we need from youNothing. This stage is ours.

Boring, proven, yours to keep.
The technology we work with across client builds, our own products and the tooling behind them. Whatever a project runs on, the accounts are yours at handover.
- Application and design
- TypeScript · Node.js · React · Next.js · Astro · Figma
- Platform and data
- Cloudflare · GitHub · Supabase · Postgres · Resend
- Models and tools
- Anthropic · OpenAI · Gemini API · Model Context Protocol (MCP)
Around the working stack
Also in use
In use behind the builds, each for what its row says. Hostinger holds domains and business mail only: none of our products is hosted on it.
- Tests and audits
- Playwright · Vitest · axe-core · Lighthouse
- Front end
- Tailwind CSS · Vite
- Documents
- PDF.js · pdf-lib
- Domains and mail
- Hostinger
Scoped on your account
Named because your business may already run on them. We have not delivered a project on any of them. Where yours needs one, the integration is written into the scope, then built and tested against your own account, within the provider’s terms and the markets it serves.
- Commerce
- Shopify · WooCommerce
- Payments
- Stripe · PayPal · Klarna · Razorpay
- CRM and email
- HubSpot · Salesforce · Klaviyo
- Domains
- GoDaddy · Namecheap · Squarespace Domains
- Sign-in
- Microsoft Entra ID · Google Workspace · Okta
- Databases
- MySQL · MongoDB · Microsoft SQL Server · Firebase
- Cloud
- AWS · Microsoft Azure · Google Cloud · Vercel
- Model hosting
- Amazon Bedrock · Microsoft Foundry · Gemini Enterprise Agent Platform
Third-party names are used descriptively. No partnership or endorsement is implied.
Standard
What ships with every build
None of that is an upsell. It is the difference between something that was generated and something that was built.
We run the audit and the checks ourselves, and hand over what was tested and what was found. We do not certify compliance with any standard or regulation.
- A written test suite. Written as the build goes, and run in full before anything goes live.
- An accessibility audit against WCAG 2.2 AA. Automated checks on every page at seven widths, and keyboard checks. It is not a claim of conformance.
- A performance budget the site has to pass. Measured with a Lighthouse run before launch. You get the numbers, not a description of them.
- Security checks before launch. Response headers, input validated by any server code we write, and secrets kept out of the repository. It is not a penetration test.
- Documentation. A written record of how it works and how to change it.
- Accounts in your name.
When the build includes a model
- An evaluation set built from your own real questions, run before handover, with the results published rather than described.
- Where the system can call tools: a written map of what each tool may read and change, a person’s approval before any change is made, and a log of every call, kept in your account.
- A written record of what the system does not cover.
Every project starts with a written scope. Ask for yours.
Kordal Systems is a design and engineering studio. We build websites, web applications and AI systems for e-commerce brands and B2B companies, and run software products of our own.
Kolkata, India · We reply within one business day.





