Dinika:

Duncan, you could have built anything coming out of GoCardless. What made you see opportunity in accounts receivable?

Duncan:

At GoCardless I spent years helping businesses get paid, but when you're in a box handling the payments you can't solve the whole problem. The real opportunity comes when you can control all the levers - not just how people pay, but how you communicate and negotiate with them. This has always been done by humans in a way that has barely changed in hundreds of years - AI is the unlock that finally makes it possible.

Dinika:

Beyond your resume, what convinced me you were the right founder for this specific problem was your customer obsession. Every conversation always came back to the customer - that came across in every interaction and still does.

You told me very early you wanted to build something AI native that would change how people work. Walk me through how you think about this as a "communication and negotiation problem" rather than a payments problem.

Duncan:

I think most people expect B2B invoicing to work like this: Company A issues an invoice, Company B pays it. If you think like that, the challenge is getting money from Company B to Company A.

Reality is quite different.

You typically see Company A attempt to identify what Company B owes; struggle to find the right people to contact about it; trade emails about what payment options are possible and whether the invoice is correct; reach out to several colleagues and departments to resolve the matter… and then ultimately when it's paid and how depends on the balance of power. I had a public company CFO tell me his AP team was under instructions not to pay the lawyers until they chased him personally. This is all about communication and negotiation - it wouldn't change at all whatever payment methods we come up with!

Dinika:

That's exactly the insight that made this compelling to us. You've called it the "Wild West" of receivables - what does that metaphor actually capture?

Duncan:

This isn't obvious from the outside, but when you truly get into the weeds and spend time with the finance people dealing with it, you realise it really is wild out there. I mean WILD.

I speak to people who have spent 10+ hours on a single invoice, going back and forth, debating it, finding documents, asking their colleagues, chasing endlessly. And it's endless. You are never done. More customers and more invoices are always piling up - it's this never-ending task of trying to herd cattle and tame those wilds.

Dinika:

Many investors and founders are chasing the sexy parts of fintech. But from our perspective, the most acute problems are probably the least sexy to solve. Through your journey at GoCardless it was clear it's not always super easy to build something sexy in payments - you need to clean the pipes before you can build around it. Our belief is if you can solve unsexy problems, you can build on top of it and turn it into something sexy.

Why is now the right moment for this approach?

Duncan:

One of the biggest opportunities AI presents is the automation of work, and AR is a place that involves deeply disliked work - no one likes chasing after invoices, but it's a human, very intensive, manual process. Every human interaction is involved in chasing invoices and receipts. Automating it could be a magical experience if it works.

Dinika:

When you look ahead - what does the world look like if Lunos succeeds?

Duncan:

If Lunos succeeds, the back-and-forth of AR and AP won't drag on for weeks - it'll be resolved instantly by agents acting on behalf of businesses. Payments, disputes, and negotiations will all move at machine speed. As those agents connect, cash flow becomes fluid and tradeable across the network, no longer locked inside slow, manual processes.

That means businesses don't fail just because they can't get paid, and the economy as a whole grows faster.

Dinika:

From Cherry's perspective, we believe Lunos will be an end-to-end B2B agentic payments network, with receivables just as a starting wedge. When you talk about agent-to-agent commerce, that's not just about AR - that's about reimagining how all business gets done.

Final question - everyone's building "agents" now. What separates real AI companies from AI features?

Duncan:

You need to be obsessed with truly understanding and solving customers' problems, before leaping to solutions. Combining this deep understanding with taste and insight is how you build "cars, not faster horses." Focus can sound like a buzzword, but there's a natural tendency to try to do more and more at once and you have to fight it. It's usually better to make fewer, bigger bets.

Dinika:

Exactly. It's clear that Workday could not build Lunos. They could build a much worse version of Lunos, but not one that pulls data, understands context, and over time is able to sound and act like an actual human worker.

Audio version recreated by AI: