01"Our commission is paid by the supplier and disclosed in full, category by category."
Requires an immutable commission ledger: rates versioned per supplier and category, locked at activation, joined to transaction volume, and rendered onto a statement the operator can reconcile independently. Disclosure is only credible if it is computed, not typed.
02"Share your spend data. Receive a category-by-category savings plan."
Requires a spend intake engine: document AI over invoices, contracts and statements in whatever format arrives, line-item extraction, unit normalisation, supplier entity resolution, and mapping into a canonical category taxonomy. Today this is analyst weeks. It should be a 48-hour automated pass with humans on exceptions only.
03"We show how your unit prices compare — venue against venue, and against the wider pooled book."
Requires a benchmarking engine over a normalised pooled dataset, with unit-of-measure reconciliation (a 2kg pack and a 500g pack are not comparable until something makes them so) and privacy-preserving aggregation, so no operator can reverse-engineer another operator's pricing from the benchmark.
04"The one site paying nine percent above contracted rates. We find it, cost it, and bring it back into line."
Requires leakage detection: continuous variance monitoring of actual invoice prices against contracted rates, per venue, per SKU, with a costed exception queue and a workflow that closes the loop rather than just raising a flag.
05"Contract Integrity Monitor — flags the moment any term drifts from what was negotiated."
Requires contract intelligence: clause extraction into a structured schema at execution, then continuous comparison of live invoice and delivery behaviour against those extracted terms. This is retrieval-augmented document AI with a strict schema and a human review path — precisely the architecture we shipped for a legal-AI platform in London and a bank in New York.
06"We track every expiry across your portfolio and run the renewal ahead of time."
Requires a renewal and obligation calendar populated automatically from extracted contract dates, with notice-period arithmetic, auto-renewal trap detection, escalation clauses, and tasking that reaches a human before the window closes — not a shared spreadsheet with reminders.
07"Supplier Performance Scoring — delivery, quality, pricing consistency, disclosure compliance, in real time."
Requires a supplier scoring pipeline ingesting delivery and fill-rate signals, credit notes, complaint records and price variance, weighted per category, versioned so a score can be defended when a supplier disputes it. The number on your homepage today reads FOOD // SCORE: 94.1. Something has to compute that.
08"Your commercial information belongs to your operation. Full stop."
Requires an enforced data posture: per-tenant isolation, Australian data residency, PII redaction before any model call, enterprise model APIs under no-training clauses, full logging of every model invocation, and an access trail an owner can audit. A promise not enforced in architecture is a promise waiting to be broken by an integration.
09"The Tendia Terminal — every venue, agreement and dollar in one place. Every month, without asking."
Requires the product itself: a multi-tenant SaaS platform with role-based access across owner, venue manager and Tendia category manager, scheduled reporting, and an export path. Right now the Terminal is the strongest thing you sell and the thing most exposed if it is not there when a serious operator asks to see it.