He Built the Network. Now He’s Going Back for the Part It Couldn’t Reach.

There is a particular kind of founder we look for: the one who took a category as far as the last generation of technology allowed, watched it stop, and knows exactly where the wall is. Craig O’Neill is that founder for B2B payments.

What he already built

He became CEO of Versapay in 2013 and pivoted a payments company into collaborative accounts receivable, a category he effectively named. The pitch was simple and, at the time, unusual: the invoice-to-cash cycle involves too much paper and too much manual intervention, and fixing it requires the buyer and the seller to work inside the same digital space rather than mailing artifacts at each other.

It worked. Great Hill Partners took the business private in 2020. By January 2023, more than a million companies were transacting on the Versapay network, on the back of a record 82% year-over-year increase in software bookings. Along the way he acquired DadeSystems specifically to put machine learning underneath cash application, because even then it was evident that matching payments to invoices was a job for a machine rather than a person.

He is a computer science and mathematics graduate of the University of Toronto who spent twenty years shipping enterprise software before any of that. He is not a first-time founder discovering that finance is hard.

Where it stopped

Here is what O’Neill knows that almost nobody else building in this space does: precisely how far the last generation of software got, and precisely where it ran out.

It got to the workflow. It digitised the paper, moved the portal online, gave the AR team visibility and gave the buyer a self-service view of their own account. Enormous value, genuinely. And then it stopped, because everything past that point was judgment. The partial payment. The disputed line item. The customer whose remittance format changes quarterly. The account that pays on its own schedule and calls that a policy. Those exceptions were left to people, and they are where all the cost, all the delay and all the working capital sit.

Workflow software could not cross that line. It could route an exception to a human. It could not resolve one.

Argentic is O’Neill’s second run at the same wall, with tools that did not exist the first time. The difference between an AR platform that surfaces an exception and an agent that clears it is not incremental. It is the difference between software that watches the work and software that does it.

Why we backed the second attempt

Repeat founders are common. Repeat founders returning to the same problem are rare, and they are a distinct and better asset.

Most second-time founders arrive with a network, a reputation and a fresh domain they have to learn. O’Neill arrives with a decade of accumulated knowledge about a single market: which ERP integrations actually matter, which mid-market segments will pay, what a controller will and will not hand over to a machine, how a collections team really behaves in the last week of a quarter. That knowledge does not depreciate. It compounds, and it is only useful to someone who goes back.

He also carries the least-discussed advantage in enterprise software: he knows the incumbents from the inside, including the one he built. He is not guessing at where they are slow.

The thesis fit

Our claim is that critical global systems are constantly expiring, and that the founders best positioned to rebuild them are the ones who lived inside the last version. B2B payments is a system that processes a staggering share of global commerce on infrastructure that still assumes a human will read the email.

O’Neill built the version that worked for its era. Argentic is the version for this one.