Bilanso
Multi-tenant marketplace reconciliation SaaS for e-commerce operators running Mirakl, Amazon, and Allegro channels - live at bilanso.io.
// delivery facts
Multi-tenant SaaS
Architecture
RLS isolation per organisation, company_id scoped throughout
3
Channels
Mirakl, Amazon, Allegro - dedicated adapter per channel
Live - bilanso.io
Status
Running in production
Drizzle + Supabase
ORM
Type-safe schema with migration-based deployments
The challenge
E-commerce operators selling across multiple marketplaces (Mirakl, Amazon, Allegro) faced a reconciliation black hole: settlement files, commission invoices, and bank payouts came in different formats with no automated way to match them.
- 1
Settlement files from Mirakl, Amazon, and Allegro each use different schemas, currencies, and fee structures
- 2
Commission invoices and bank payouts had to be manually matched to settlement lines - a process taking hours per period
- 3
Multi-bank MT940 imports from different banks arrived in slightly incompatible dialects of the same format standard
- 4
No single export path to accounting systems - each operator maintained bespoke manual exports
How it was built
Multi-tenant SaaS on Next.js, Supabase, and Drizzle ORM with dedicated marketplace adapters, MT940 multi-bank parser, automated matching engine, and accounting-system export layer.
Supabase RLS with company_id-scoped policies on every table - tenant data is physically isolated, not just filtered at the application layer
Dedicated adapters for Mirakl, Amazon, and Allegro normalize channel-specific settlement schemas into a unified internal representation before matching
MT940 parser supports multiple bank dialects simultaneously; bank-specific quirks are handled per-bank without polluting the shared pipeline
Matching engine links settlement lines to commission invoices and bank payouts using order-reference exact matching with fuzzy fallback (name, amount, date proximity)
What shipped
Production multi-tenant SaaS live at bilanso.io - three-channel marketplace reconciliation, multi-bank MT940 import, and full RLS tenant isolation shipped.
// full stack
More case studies
4Fleet
Multi-tenant B2B fleet-management SaaS with an AI agent for email-to-quote parsing, dynamic pricing engine, and Stripe billing - live at 4fleet.io.
NumWise
Accounting SaaS with format-agnostic bank-statement import (CSV, XLSX, OFX, QIF, MT940, CAMT.053 from any bank or country), AI-assisted column mapping, and 113 shipped features - live at numwise.eu.
Want something built like this?
Free qualifying call - 20 minutes. Fixed price. Real code. You own it.