AmuraAMURA Software
Service · AI code audit · Distribution & B2B industry

AI code audit for distribution and B2B industry.

You shipped an AI quoting tool on Cursor, a customer portal on v0 or an order assistant on Lovable. We audit before one customer sees what you priced for another — or before an ERP token leaks.

What we solve

In B2B, a leak isn’t a bug — it’s lost margin.

Every B2B customer has their price, their discount, their payment terms, their reserved stock. The AI tools your team built (quoting tools, customer portals, order assistants) handle all of that. When the ownership filter fails, Customer Small can see the price you charge Customer Large. An external audit catches it before your sales team does.

We audit what your team built through the sector lens: strict isolation between customer accounts, custody of ERP tokens, exposure of cost and margin data, and the spots where an AI agent has more permissions on your ERP than it should.

What we build for this sector

Use cases that ship to production.

See full catalogue →
Isolation

Per-customer pricing and terms isolation

Each customer sees only their prices, discounts, payment terms. We verify at the SQL layer — not just the UI — and across the public APIs your quoting tool or portal exposes.

0 cross-account price leaks
ERP

AI agent permissions on the ERP

Tokens to SAP, Holded, Odoo, Sage, Microsoft Dynamics. What the agent can read, what it can write, what it should never touch (chart of accounts, master agreements, vendor data). Explicit deny.

Audit-trail per write
Cost & margin

Cost, margin and stock data that shouldn't leave

Quoting endpoints returning internal stock, vendor cost or computed margin in responses customers can inspect. What the UI hides, the API sometimes reveals.

Sensitive fields out of responses
Multi-tenant

Isolation in multi-entity setups

If your group runs multiple legal entities, brands or channels (wholesale vs. retail), we audit that a Channel-B customer can't reach Channel-A data by changing a subdomain or tenant ID.

Tenant boundary verified in code
A real scenario

A distributor with 1,800 active accounts.

Industrial-supply distributor with an AI quoting tool built on Cursor over PostgreSQL plus a customer portal on v0 over Supabase. Audit done after an internal incident where two customers saw prices that weren't theirs.
Before the audit

The quoting tool worked: each customer signed in with their email and saw their catalogue. The price query, however, was reading the full catalogue per customer_id and filtering in the UI. Any customer with DevTools could see every price — including those of the customer who buys 10× the volume. The Holded ERP integration used a token with write permissions on chart-of-accounts entries the agent should never have touched.

After the audit

14 findings. 4 criticals: ownership filter moved to SQL, quoting responses cleaned of cost and margin, Holded token replaced with a minimum-permission one, log policy that had been letting full orders into Vercel logs. The remaining 10 findings documented with deadline and owner.

4 criticals resolved before the next contract renewal cycle
We connect to your stack

Integrations that matter in this sector

CRM

Salesforce

Enterprise CRM with fine-grained permissions — AI workflows that respect the data model.

CRM

Microsoft Dynamics 365

Enterprise CRM/ERP suite in the Microsoft ecosystem — native fit with 365 and Power Platform.

ERP

SAP Business One

Reference ERP for mid-market distribution and manufacturing — document extraction and ops orchestration.

ERP

Holded

Spanish cloud ERP widely adopted by SMBs — invoicing, expenses and reconciliation automation.

ERP

Odoo

Modular open-source ERP — AI agents and workflows on top of sales, inventory and project modules.

COMMS

Microsoft 365 / Outlook

Email, calendar and SharePoint as channel and context — triage, drafting and RAG over your inbox and files.

Frequently asked

What clients ask us

  • 01

    We handle cost, margin and per-account terms. How do you treat it during the audit?

    Under NDA, with read-only repository access. We don't copy real commercial data. If the audit needs to probe a real flow, we use test accounts your team sets up in staging.

  • 02

    Our ERP is SAP / Holded / Odoo / Dynamics and the AI agent writes to it. Do you cover that?

    Yes. We audit the connector: which token it uses, which objects it can read, which it can write, how it tells apart environments (sandbox vs production), how it handles errors and retries. If the connector writes to chart-of-accounts or stock, we read closely what stops it from writing when it shouldn't.

  • 03

    Can you detect whether a customer is seeing another's prices right now?

    We audit the quoting flows and API responses to find price leaks. It isn't real-time monitoring — that's a separate service — but the report tells you exactly what to change so the leak isn't possible and how to verify it hasn't happened historically.

  • 04

    We run multiple channels (wholesale, retail, white-label). Do you audit isolation between them?

    Yes — and it's usually where the most findings cluster. B2B multi-tenant mixes legal entities, subdomains, shared tables and overlapping pricing rules. We map every boundary and verify the code respects it, not just the documentation.

Trust

Safe, traceable AI,
enterprise-ready.

We design for privacy from the start, human control, traceability, usage limits, permissioning and documentation. For sensitive processes, we help assess risk and applicable obligations under GDPR and the EU AI Act.

  • 01We never train models on your data without explicit authorization.
  • 02Human review built-in for processes where risk demands it.
  • 03Traceability: prompts, sources, permissions, errors and metrics — documented.
  • 04Privacy, security and control integrated from day one.
  • 05Solutions engineered to be maintained, audited and improved over time.
GDPREU AI ActAEPDISO 27001 readyEU data residency
Personal diagnosis

We work with
few clients.

Every engagement is led personally by one of the partners. If there's a fit, you get a personal first read of your case within one business day — not a canned demo.

How we work
  1. 01Tell us which process eats your time
  2. 02Personal reply within one business day
  3. 0320-minute call — no demo, no pitch
Start the conversation →