BPSI France — Base44 → Supabase migration

→ Business impact: Migration without downtime + removal of the proprietary-vendor lock-in — lower operating cost and full control of the stack regained.

Context

BPSI France runs office fit-out projects across several European offices. Their platform (marketing site + careers + case studies + client portal) ran on Base44, a proprietary backend: strong dependency, little control over the data, growing cost and lock-in.

Goal: take full control of the stack — data, auth, emails, deploy — without ever taking the production site or a company mailbox down.

Technical challenge

113 Base44 references scattered across the app to remove entirely.
Reconstruct the data model (jobs, case studies, profiles) on Postgres with clean access policies (RLS).
Migrate with zero downtime: the old site had to keep serving throughout.
Email trap: the domain carries active mailboxes (Titan) — a bad DNS move would have wiped them.

Solution

Data model rebuilt: 7 Postgres tables + Row Level Security, real data re-imported (open jobs, case studies, profiles).
8 Edge Functions (Supabase) for the API, sitemap, assets, and email delivery.
Quote form wired to Resend (transactional email).
Front-end Vite + React deployed statically on Cloudflare Pages.
Progressive DNS cutover (A/CNAME records surgically edited, MX/SPF/TXT untouched) → no email lost, old site as a safety net, zero downtime.

Outcome

113 Base44 references → 0. Full exit from the proprietary backend.
Migration delivered in 8 clean commits, staging verified end-to-end.
Corporate mailboxes preserved (Titan risk documented, no mailbox touched).
A written deploy handoff: commands, instant rollback plan, checklist.

Learnings & trade-offs

The real risk in a migration isn’t the code, it’s DNS and email. Most of the care went into a reversible cutover, not the refactor.
Document every irreversible step before executing it — and know when to stop in front of a dangerous checkbox.