Your next ten clients are already in your data

Most funders sit on an origination asset they have never opened. It is not a dataset they bought, and it is not a lead list. It is the paperwork of their own book: every invoice, every contract, every KYC file names companies, and those companies are connected to each other in ways the documents record but the filing system forgets.

SQF does not forget. Every document a funder processes on the platform adds companies and relationships to a private knowledge graph — a live map of the trade network around that funder's book, visible to no one else. The lending happens as it always did. The graph does the remembering.

A map drawn from your own paperwork

The graph builds itself from ordinary work. When an invoice names a counterparty the funder has never met, SQF records the company automatically. An AI analyst screens it the moment it appears — compliance check and credit report — and a human confirms the finding before it stands. No unvetted company sits silently in the map.

Agency KYC files contribute the human layer. Directors and shareholders named in those files become people in the graph, connected across companies: director-of, shareholder-of, has-interest-in. A person who appears in two unrelated files becomes one node with two edges, and a connection nobody typed into any system becomes a fact you can query.

The result compounds. More flow makes a richer graph; a richer graph surfaces more origination; more origination brings more flow. What follows are four worked examples from real demonstrations of the platform.

Two origination playbooks

Anchor-led supplier finance. Your client Meridian Foods buys from 14 suppliers you have already seen on invoices. Nine are not your clients yet — but they are KYC-screened, and their volumes and payment terms are on record: the top three invoice RM 4.2M, RM 2.9M and RM 1.7M a year at 60-day terms. One query returns the list, ranked. That is an anchor-approved early-payment programme with nine pre-qualified prospects, sourced entirely from your own flow.

The factoring cross-sell you already underwrote. A supplier invoices three separate buyers in your network, and you fund only one of those relationships. The graph shows their whole receivables position across your book: RM 6.1M a year at 75-day average terms, with payment behaviour you have already observed. That is an invoice-finance conversation where the underwriting evidence exists before the first meeting — the rarest thing in origination.

Risk you can see three hops away

The graph works in the other direction too. Consider two borrowers with no visible connection between them. The graph knows what the credit files, read separately, do not show: the same person is a director of one and a 55% shareholder of the other — and both are supplied by a third company in which he holds an interest.

That is three hops, resolved in one query, from edges built out of agency KYC files and filtered to current officers only. It puts related-party concentration in front of the credit committee before the decision, rather than in front of a workout team after the default. Concentration you can see is concentration you can price. Concentration you cannot see is a loss provision waiting for its trigger.

The growth signal

Not every finding is a new name. Sometimes the graph and the ledger, read together, say something about a client you already have. A client's invoice throughput grows 40% across two quarters while facility utilisation holds above 90%. Each figure alone is unremarkable; together they describe a working-capital gap — a business growing faster than its facility.

The platform flags it, and your relationship manager opens the term-loan conversation before the client starts shopping for one. The alternative, in most books, is that the first sign of the gap is the client's polite email announcing a refinancing elsewhere.

Ask it in plain language

None of this requires a query language. A funder can ask the graph a question in plain words — "Which companies in my network supply more than one of my clients?" — and the answer comes back as companies, connection counts, and the evidence trail behind each one: which invoices, which files, which confirmed screenings.

A set of saved opportunity queries ships with the platform, covering the playbooks above. Beyond those, natural-language questions are compiled to graph queries on demand, so the analyst asking the question does not need to know how the map is stored — only what they want to know about their own network. Every answer carries its sources, because an origination lead a relationship manager cannot verify is a lead they rightly will not act on.

Whose data this is

One boundary is worth stating plainly. Each funder's graph is its own: built only from that funder's flow, held in a dedicated environment, and never pooled, shared, or used to inform anyone else's map. The personal identifiers used to resolve one person across several companies are used for exactly that and are never exposed through any interface. The competitive asset the graph represents belongs to the funder whose paperwork built it.

That is the whole argument. A trade-finance book generates, as a by-product of its ordinary operation, a screened and evidenced map of the commercial network around it. Funders have always owned this asset. SQF makes it legible — and the next ten clients tend to be on the first page.