This deep dive expands on the Loop Asia conversation with Fethi Rabhi, Professor of Software Engineering at UNSW, and Alan Hsiao, founder and CEO of Cognitivo.

Alan Hsiao dropped a number that deserved more time than the format gave it. Invoice factoring — a supplier sells unpaid receivables to a bank for cash now instead of waiting 30 or 60 days — typically costs around 5% off face value on a 30-day invoice. Compound that and you're at roughly 27% APR: a striking cost of capital most executives never notice, sitting inside their own supply chains.

Alan's explanation isn't opportunism — it's process cost. Banks offering factoring manually work out each invoice's due date, verify the supplier, check the line items: "so the bank is doing the invoice processing," as he put it. A thin-margin product becomes, once you price in the manual handling, one of the more profitable things a bank does. The inefficiency isn't a side effect — it's the reason the price is what it is.

Why this is your problem even if you never touch factoring

If you run professional services delivery, or own a P&L that depends on supplier or customer payment cycles, you don't need to use factoring for its economics to matter. If your suppliers finance their receivables — and in a just-in-time, thin-margin chain, a meaningful share do — that 27% cost is baked into their pricing, their willingness to extend terms, and their resilience when something breaks: an invisible line on someone else's balance sheet that shows up as risk on yours.

The rate is high, per both guests, because of trust, not technology. PEPPOL solves transmission — sent, received, not corrupted — but not who can rely on an invoice's current, authoritative state through approval, dispute, and payment. Alan's fix: a shared, immutable record — distributed ledger, one of the rare cases he thinks it's genuinely the right tool — a factoring bank could rely on without re-verifying by hand: "we don't have the double-spending problem in this case," because everyone's looking at the same copy. Whether blockchain wins or not, the argument holds: the expensive part of invoice financing is verification, not capital.

The monolithic shortcut, and why it doesn't fix this

Large enterprises already "solved" this trust problem — badly. A big buyer runs a procurement platform like Coupa: every supplier logs in and re-enters their invoice on top of the one already generated in their own accounting system. It gives the buyer a trusted record; it does nothing for the inefficiency, just relocates the manual labor onto every supplier, for every relationship. Alan's assessment: "extremely, extremely inefficient... it's not really solving the problem at all."

This matters if you're a Head of Professional Services or COO under pressure to stand up an integration fast after closing a large client. Requiring every partner to conform to your platform looks like control; it's a tax on every relationship, compounding with each one you onboard. The alternative Alan and Fethi pointed toward — peer-to-peer exchange over PEPPOL, with a trusted shared record — costs more up front and less at scale.

What this means for the next 3–5 years

Neither guest thinks agent-mediated settlement is imminent — Fethi was explicit that most companies are still absorbing basic API and XML adoption, let alone autonomous agents. But the direction Alan sketched — agents that read an invoice, verify it against a shared record, understand the law, and chase or release payment automatically — answers the 27% problem rather than speculating around it. The economics already justify solving it.

If growth is gated by how fast you can extend trustworthy payment terms or absorb new suppliers into a fast-moving rollout, that's the cost structure you live with for years once this quarter's platform choices lock in under pressure. Worth an outside, vendor-neutral look at the design before that happens.