Home/Database/Backend & infrastructure/Supabase Agency
Supabase Agency — Paris

Supabase as a production backend —not just as a Firebase replacement.

Scroll uses Supabase for nearly all its projects: business applications, SaaS, Next.js websites, AI tools. We know the platform’s strengths, its blind spots, and what needs to be configured correctly from the start to avoid reworking everything in production.

Supabase stackproduction backend
FrontendNext.js · React · TS
API · SDK
SupabasePostgreSQL
PostgreSQLdata · SQL
AuthJWT · RLS · SSO
Storagefiles · assets
Edge Functionsserverless
Realtimelive · pgvector
RLS per tableMigrations versioned
01 — Real-world use cases

Supabase in our projects —
real-world use cases.

No demo. Six configurations we actually deploy, from custom application backends to sovereign hosting — each with its own specific requirements.

01

Custom application backend

PostgreSQL schema designed for the use case, versioned migrations, Row Level Security configured from the start. Not an auto-generated schema that needs fixing later.

02

Authentication and role management

Supabase Auth with JWT, fine-grained role and permission management via RLS, SSO possible (SAML, OIDC for enterprise contexts). Multi-provider authentication configured and tested.

03

Backend for AI applications

Vector embeddings storage via pgvector, API for LLM agents, MCP servers that expose Supabase data to the model in a controlled way. The database as the AI’s memory.

04

Firebase replacement

Migration from Firebase (Firestore, Auth, Storage) to Supabase — relational schema rethought, data exported and restructured, authentication migrated. Zero downtime.

05

Airtable replacement

For teams that have outgrown Airtable: migrating bases to PostgreSQL, recreating automations in n8n, building a custom business interface if needed.

06

Sovereign hosting

Self-hosted Supabase on OVH or Scaleway for contexts where data cannot leave controlled infrastructure (healthcare, finance, sensitive data).

02 — What we address from the start

The topics we address from the start —
not as an emergency in production.

"Doing Supabase" and "building a production-ready Supabase backend" are not the same thing. The difference lies in four key areas we address during design, not after an incident.

Row Level Security

Supabase’s RLS is powerful—and dangerous if misconfigured. A missing or overly permissive RLS policy exposes data. We define and test RLS policies for each table from the schema design phase.

Scalable data schema

A poorly designed schema from the start is costly to fix once the app is in production. We design the schema with future evolutions in mind, using versioned migrations via Supabase’s migration tool.

Performance under load

PostgreSQL is performant, and so is Supabase—if indexes are properly set and queries are well-written. We audit slow queries, identify N+1 issues, and configure connection pooling before it becomes a problem.

Backups and recovery plan

Supabase Cloud offers automatic backups. In self-hosted setups, this must be explicitly configured. We define the backup and disaster recovery strategy from the start, not after an incident.

They trusted usSee our client cases
Ubki
Perfway
Hexa
Art Explora
Bellman
Cabaia
03 — In the stack

How Supabase integrates
into a Scroll stack.

Supabase provides the core — data, auth, storage, serverless, realtime, vectors. Around it, open-source building blocks complete the architecture: frontend, back-office, automations, sovereign infra.

LayerToolRole
DatabaseSupabase (PostgreSQL)Source of truth, SQL queries, migrations
AuthenticationSupabase AuthJWT, RLS, SSO, multi-providers
File storageSupabase StorageAssets, documents, images linked to data
Serverless logicSupabase Edge FunctionsWebhooks, triggers, lightweight processing
RealtimeSupabase RealtimeLive updates, notifications, collaboration
AI & vectorspgvector on SupabaseEmbeddings, semantic search, RAG
FrontendNext.js, React, TypeScriptInterface connected to the Supabase API
Admin / back-officeDirectus or custom interfaceNon-technical access to data
Automationsn8n self-hostedWorkflows connected to the Supabase API
Self-hosted infrastructureOVH, Scaleway, DockerSovereignty if required
Supabase platformSurrounding components
04 — Hosting

Cloud or self-hosted —
how to choose.

Both are genuine Supabase. The choice comes down to one real criterion: can your data be hosted on AWS EU, or must it remain on an infrastructure you control?

Supabase Cloud

EU region

For most projects: quick start, automatic backups, managed updates, no ops to handle. Data is hosted on AWS eu-west (Ireland) — GDPR-compliant for most use cases.

We recommend when

No strict sovereignty constraints, no dedicated ops team, priority on speed to production.

Supabase self-hosted

OVH / Scaleway

For projects where data cannot leave a controlled infrastructure: healthcare, finance, data covered by demanding client contracts, public sector. We deploy and maintain the Supabase infrastructure on OVH or Scaleway.

We recommend when

Sensitive data, contractual location requirements, contexts where AWS is not acceptable.

05 — FAQ

Frequently asked questions

The most common questions we get during scoping. If yours isn’t here, reach out!

Firebase (Firestore) is a NoSQL database—flexible for getting started, but limiting for complex queries and joins. Supabase is PostgreSQL: relational, full SQL support, and far better suited as data logic grows more complex. Plus, PostgreSQL doesn’t lock you into Google’s ecosystem.

Xano is a no-code BaaS with a visual interface. Supabase is an infrastructure that developers configure with code. If you have a technical team, Supabase offers more control, lower costs, and full portability. If you don’t have developers, the question changes entirely.

Payload has its own database layer (MongoDB or PostgreSQL, depending on the config). Directus can connect to an existing PostgreSQL database—so it works with Supabase. We choose the architecture based on the project: Supabase alone for a custom app, Directus + Supabase for auto-generated admin needs.

Yes—it’s actually one of its current strengths. The pgvector extension lets you store embeddings directly in PostgreSQL and perform semantic search. We build RAGs and AI agents with Supabase as the memory and data layer.

Scroll can handle maintenance (updates, monitoring, backups, performance management). We can also transfer responsibility to your team—full infrastructure access is yours from the start.

Yes. Supabase is PostgreSQL. A standard export is all you need to migrate to any PostgreSQL infrastructure—RDS, Neon, dedicated instance. No proprietary formats, no dependency on a closed API.
Get started

Supabase is in your stack or you're evaluating it — we can review it together.

Technical scoping, existing configuration audit, or starting from scratch: we adapt to where you are.

Contact details
contact@agence-scroll.com
+33 6 48 03 90 27
20 Rue des Taillandiers
75011 Paris
Response within 24 business hours.