George Kurdin learned when to fold. Then he built Monk.

George Kurdin named a company after Gattaca, the film with a swimming scene about refusing to give up. Then he gave up on the company. It was profitable. It had a product, built around computer vision for quality assurance. He returned the remaining investor capital anyway. The market was too small, a co-founder partnership had ended, and persistence had stopped looking like a useful answer.

That is an awkward opening for a founder biography. The customary version includes a garage, a rejection, and the eventual humiliation of everyone who said no. Kurdin supplies a more interesting decision: recognizing when the next hour belongs somewhere else. The swimming scene remained a good scene. It was a poor substitute for a large enough market.

Today he is the co-founder and CEO of Monk, a New York company building AI software to help businesses collect money they are already owed. The distance from cinematic determination to overdue invoices is considerable. His route between them runs through professional poker, finance, creator software, and a partnership that began with conversations about life while training for a marathon.

The education of a poker player

Kurdin spent six years playing poker professionally, logging more than 1.5 million hands. One session lasted about 60 hours. Those numbers suggest a man unusually willing to remain seated. The lesson he carried away concerned movement: adjusting decisions quickly when the game changes.

A good decision can lose a hand. A bad one can win. Repeat that experience often enough and the temptation to judge your intelligence by the latest result becomes expensive. Kurdin describes poker as an education in risk, expected value, and discipline. The familiar movie version, with its microscopic reading of an opponent's twitching eyebrow, has rather less to do with his account.

There is a useful tension here. A founder needs confidence to make decisions before the outcome is known. Confidence also makes it easy to defend a decision long after the reasons for it have expired. Poker offers frequent, costly opportunities to notice the difference.

“It's not that sexy, but it's very important for the world.”

George Kurdin, on accounts receivable

A career in the details people use

His working life has included D.E. Shaw, Minecraft, and Streamlabs. He credits the finance experience with teaching precision and the software businesses with teaching him to ship products and run teams. Together, those are useful qualifications for a company whose software must interpret an email and also respect an accounting record.

The Streamlabs years leave a particularly tangible trail. His published articles discuss overlays, stream chat, simulcasting, and recording video locally. A January 2022 piece explains how separate local recordings can improve audio and video quality regardless of a participant's connection. These are specific, everyday frustrations for people trying to make something in public.

There is continuity between helping a broadcaster preserve a recording and helping a finance team match a payment. Both require attention to the uncooperative detail that interrupts the promised experience. A product can look perfectly convincing in a demonstration while its user spends the afternoon repairing the handoff.

His personal newsletter, ace high, offers another clue to his manner. Its stated subjects are building, product, startups, and summaries of what he reads. He introduces himself as a “Mediocre poker pro turned product/ops.” The biography acknowledges mistakes as readily as accomplishments. The title keeps the old card table in view.

The company after the company

After Gattaca, the next partnership came through South Park Commons. Kurdin and Joe Zhou were introduced there; someone thought their intensity might be compatible. They spent time together and trained for a marathon before deciding to start a business. Choosing a co-founder began with getting to know a person, rather than dividing titles on a slide.

Monk took shape around a problem finance departments could recognize without much persuasion. A sale creates an invoice. An invoice creates a claim on money. Between the claim and the bank deposit, a surprising amount of human coordination can intervene. Someone needs a document. Someone else needs an approval. A payment arrives without enough information to match it.

In his August 2026 account, Kurdin said the founders spoke with more than 100 CFOs before starting Monk. The recurring issue was not a shortage of software. Teams already had tools. Billing, communication, and reconciliation still failed to join into a dependable whole.

That distinction gave the work a concrete destination. An additional dashboard might make the delay easier to observe. Finishing the outstanding administrative work could shorten it. Kurdin's new business would have to earn its place somewhere between the invoice and the money.

A sale with very little ceremony

One early customer kept emailing. Kurdin could not make a call, so he sent a one-page explanation and a Loom walkthrough. The customer asked for the paperwork. The pilot was worth $36,000. A transaction of that size usually comes with more calendar invitations.

The episode is useful because it makes the buyer visible. Someone had a sufficiently pressing problem to keep following up with the person selling the answer. Monk later reported reaching $1 million in annual recurring revenue in roughly six months, using cold outbound without a sales team. Those are the company's account of its early progress, rather than a timetable another founder can order.

Kurdin's later reflections on sales are less breezy than the pilot story might suggest. By October, he was emphasizing how much sales and marketing matter, revising his earlier faith that building a good product would bring customers by itself. The engineer had acquired an appreciation for distribution. A product still needs to reach the person whose afternoon it might rescue.

When the machine needs permission

Monk works across invoicing, collections, payment matching, and disputes. Its collections agent, Julia, operates inside workflows and rules. The company describes testing model calls against thousands of unusual cases and surrounding them with deterministic code. In plain English, the language model has boundaries that its talent for conversation cannot waive.

That design suits the stakes. A follow-up about money is part of a customer relationship. An incorrect amount, an ill-timed reminder, or a payment recorded twice can create work far beyond the original invoice. Fluency is helpful; the permission to act requires more careful construction.

Kurdin says customers can keep people involved during the first weeks and grant more autonomy as trust develops. The software is meant to reflect the customer's communication style and account history. His August explanation also sets a limit: an agent does not independently move money or change an invoice. The boundary belongs alongside the capability in any account of what he is building.

Five minutes before the pitch

In March 2026, investor Nikhil Basu Trivedi took a call with Kurdin after hurrying through airport security. Another portfolio matter had made him late. Kurdin spent the first five minutes asking about it. The conversation about Monk came afterward.

What followed moved quickly. Basu Trivedi's partner met Kurdin in New York the next day. Further meetings followed in San Francisco, and the investors made an offer by the end of the week. Basu Trivedi's recollection emphasizes the clarity of Kurdin's explanation and his conviction about the business. Even a fast fundraising story has room for an opening question about somebody else's day.

On April 21, Monk announced a $25 million Series A co-led by Footwork and Acrew Capital, with BTV participating again. Total announced funding reached $29 million. The financing was designated for research and development and further accounts-receivable products. Capital gave the team more room to pursue a broader B2B revenue platform, the ambition behind its initial collections work.

The meeting that earns its place

Kurdin's operating preferences are unusually concrete. Monk has described running with a single weekly company meeting and no engineering standups. In October, he discussed keeping work visible in shared Slack channels and writing prompt, blameless retrospectives. Ownership is supposed to travel with the work, so the person responsible can make a decision and ship.

His hiring views are similarly demanding. He worries about a merely adequate hire whose strengths make the weaknesses easy to tolerate. Judgment and the ability to learn matter heavily in his account. There is a recognizable link to poker: the quality of a decision has to survive examination beyond its most flattering moment.

By October, Monk had assigned an engineer to evaluation full-time, and Kurdin thought he had made that hire too late. The team maintained direct Slack contact with customers and received immediate feedback when an email sounded rude. Reliability, in this business, includes knowing when the software has acquired an unfortunate tone.

That brings the story back to judgment. Kurdin has stopped one company, reconsidered his faith in product alone, and built another around limits on what a machine may do. His ambition is large. Its daily test is small enough to fit in an inbox: the right request, to the right customer, with the right understanding of what has already happened.