Use Cases
Backoffices
AI-first ops — personas, assets, billing, audit, deploys.
The layer most consultancies forget: a real backoffice. Orgs, personas, assets, leads, billing, audit, deploys — unified, multi-tenant, and built so every agent action gets logged before it ships.
- Unified personas, assets, leads, billing
- Audit log over every agent action
- Multi-tenant from day one
Why
Front-of-house AI is easy; the backoffice is where projects die. Wiring the boring parts (auth, audit, billing, asset library) first is what lets the agents ship into production without a human reconciliation step.
How
- BOAI / AIBO stack as the spine — Supabase + Mongo + audit log
- Personas + role-based agent access from day one
- Every agent action logged; every output reviewable
Proof
- Backoffice apps shipped
- BOAI · AIBO · BO28
- Audit coverage
- 100% of agent runs
- Tenancy model
- multi-tenant from v1
Backoffice — AI-first ops on one spine
Inbound · Agents · Stores · Audit
Hover or tap a node to see details.
FAQ
- Why two data stores instead of one?
- Supabase (Postgres) for typed, queryable, RLS-protected records — auth, orgs, billing. MongoDB for agent outputs and media metadata where the shape varies. Right tool per layer.
- Is the audit log actually compliance-ready?
- Yes — every agent action captures input, decision, tool calls, output, and metadata. Schema's been used in regulated client engagements; survives legal review.
- Multi-tenant from day one?
- Always. Row-level security on Supabase, org-scoped IDs everywhere on Mongo. Adding tenancy later is the most expensive refactor I know of, so we never skip it.
In production
- BOAI — multi-tenant from v1
Per-org isolation on Supabase RLS + Mongo org-scoping. Audit log compliant by default.
See it - bo28 — fleet backoffice
Centralised personas, assets, leads, billing across the 13-site fleet. One UI, many tenants.
- Legal-grade audit log
Schema survived regulated-client legal review. Every action, every output, every retry captured.