All posts
July 1, 20262 min read

From GoDaddy page builder to Azure: rebuilding rawrtech.ai

AzureIaCMeta

We tell clients that their AI systems should live on infrastructure they own and can rebuild from code. So when our own website was still sitting on a GoDaddy page builder, the irony was hard to ignore. This is the story of eating our own cooking.

What we started with

A template site on GoDaddy's builder: no version control, no deploy pipeline, DNS and email tangled into the registrar's bundle, and renewal pricing that crept up every year. Fine for a placeholder. Not fine for a company whose pitch is production-grade engineering.

The target architecture

  • Azure App Service (Linux) for the site itself — the cheapest tier that supports custom domains with free managed certificates.
  • Cloudflare as registrar and DNS — at-cost domain renewals and a clean API for automation.
  • Terraform as the single source of truth for both clouds: the web app, the DNS records, the TLS bindings, and — critically — the Microsoft 365 email records that must not break mid-migration.
  • GitHub Actions with OIDC for deploys — federated credentials, so there are no cloud passwords stored in CI at all.

What we learned

Move email records first, verify, then move the website. The scariest part of any DNS migration isn't the website — it's silently breaking MX and SPF records. We imported the full record set into Terraform, diffed it against what GoDaddy was serving, and only then flipped nameservers.

Keep the apex DNS-only until the certificate validates. Azure's managed certificates need to verify domain ownership; putting Cloudflare's proxy in front too early causes a validation loop that looks like a mysterious outage.

Infrastructure as code pays off immediately. Two weeks after the migration we wanted a second site for a dev project. It was a module reuse and one tofu apply — twenty minutes, most of it waiting on DNS.

Why this matters to clients

Nothing here is exotic. That's the point: boring, versioned, reproducible infrastructure is the substrate that everything else — including every AI system we build — sits on. If a consultancy can't run its own site this way, it has no business wiring your production systems.

Next up on our own roadmap: the site you're reading this on — rebuilt as a modern app with a real blog pipeline, deployed through the same Terraform plumbing. Cooking, eaten.

Building something like this?

We help teams take AI from idea to production. Tell us what you're working on.

Get in touch