Audit SaaS récurrent, doublons et shadow IT

# Recurring SaaS, duplicate, and shadow IT audit

## Goal

Give a Procurement lead a periodic view of an organisation's SaaS fleet: which recurring expenses weigh the most, which suppliers are likely duplicates (the same vendor registered under several names), and where AI tool spend concentrates without central coordination (a shadow IT signal). Typical use case: quarterly SaaS fleet review, prep for a renewal negotiation, or a one-off audit before a budget close.

---

## MCP scope — what's visible

> ⚠️ The Spendesk MCP has no built-in duplicate detection. Matching similar supplier names is Claude reasoning over the results of `spendesk_analyze_spend` / `get_suppliers` / `list_cards`, not a native capability of the tool. If a duplicate is confirmed, merging it stays a manual action in Spendesk — no MCP write tool covers this today.

| Need | Accessible? | Via which tool |
|-------|:----------:|---------------|
| Recurring spend ranking by supplier | ✅ | `spendesk_analyze_spend` (groupBy=supplier, payableTypes=subscription) |
| Detail on a suspected duplicate supplier | ✅ | `get_suppliers` (`search` parameter) |
| Cardholder + cost center per active subscription | ✅ | `list_cards` (type=subscription, status=Active) |
| Automatic merge of two supplier records | ❌ | No write tool covers this need today |

## Instructions

1. Call `spendesk_analyze_spend` with `groupBy: ["supplier"]` and `filters.payableTypes: ["subscription"]` over the requested period (365-day max per call, split if longer) — returns the recurring spend ranking by supplier, largest to smallest.
2. In that ranking, spot supplier names that look alike (case variants, typos, different legal names for the same vendor). Confirm each suspicion with `get_suppliers` (`search` parameter) to check whether they are indeed separate supplier records.
3. Call `list_cards` with `type: "subscription"` and `status: "Active"` to see who holds each active subscription and on which cost center — several different cardholders with equivalent subscriptions opened with no visible coordination is the shadow IT signal.
4. To isolate AI spend specifically, filter the suppliers from steps 1 and 3 whose name matches known AI tool vendors and group them separately.
5. Present three views: (a) recurring supplier spend ranking, (b) potentially duplicate suppliers with their respective cardholders and cost centers, (c) "AI tools" cluster with each owner and cost center.

**Example user prompts:**
"What are my biggest recurring SaaS subscriptions and are there any duplicates?"
"Give me a snapshot of AI tool spend across the company."

<!-- meta: score=87, date=2026-07-10, source=test-results/2026-07-10-audit-saas-recurrent-doublons-shadow-it.md -->