Live from Singapore Fintech Festival, we sat down with Fluid CEO and co-founder Trasy Lou Walsh on the future of B2B payments and finance they are building with AI agents on their platform.

This interview was recorded live at Media Lounge in Singapore Fintech Festival 2025.

Timestamps

(00:00) Catching up with Trasy on Fluid since our last call in 2024;

(02:52) Understanding the automation needs for SME payments from a process perspective;

(04:26) Use cases for AI agents in the B2B payments flow;

(06:28) Learnings from driving AI agent adoption for SMEs so far;

(07:43) Biggest challenge so far driving AI agent adoption for SMEs;

(09:49) Competitive advantage for Fluid with its development of an AI agent stack;

Directed by Paulo Joquiño

Produced by Paulo Joquiño

Follow us on LinkedIn for more updates

The content of this podcast is for informational purposes only, should not be taken as legal, tax, or business advice or be used to evaluate any investment or security, and is not directed at any investors or potential investors in any ⁠⁠⁠⁠⁠⁠Insignia Ventures⁠⁠⁠⁠⁠⁠ fund. Any and all opinions shared in this episode are solely personal thoughts and reflections of the guest and the host.

Transcript

Catching up with Trasy on Fluid since our last call in 2024

Paulo: Welcome everyone to this special impromptu conversation here at the Singapore FinTech Festival. I was glad to meet a lot of our portfolio companies, but especially Trasy here from Fluid. Fluid is one of our FinTech AI portfolio companies, and they are here with an exhibit at the Singapore FinTech Festival. I’m happy to catch up with Trasy. I think the last time we spoke was around last year, just around the time that we first partnered with you guys in that round. And now I would love to get a catch-up on what’s been happening with Fluid.

Trasy: I think the last time we chatted was almost a year ago. So many things happened after. Of course, with the investment, we were able to actually grow over 10 times in the past year—the team size, revenue, and also our reach. Now, Fluid is working with 3,000 businesses in Singapore and Malaysia, and we have 2,000 transactions across Singapore and Malaysia.

Understanding the automation needs for SME payments from a process perspective

Paulo: For those who may not be as familiar, maybe we can go back in time and start with what was the initial Go-To-Market (GTM) or the initial product, and then what differences have been made since?

Trasy: Yeah, definitely. So when we first started Fluid, we positioned ourselves as a B2B lending product. But when we learned more about our clients and what they need, we realized the opportunity is a lot bigger than just B2B financing. So now we actually have a product with features to really manage the whole B2B payment cycle to make it fluid, not just for the financing.

Maybe I could give an example. One of our clients, John, runs a seafood distributor business. He delivers seafood daily, and his business is growing. He was serving hundreds of restaurants and hotels. But in the afternoon, instead of celebrating the sales, he actually had to manually key in invoices one by one, chasing payment on WhatsApp.

Paulo: Lots of paperwork.

Trasy: Lots of manual work, lots of repetitive work, and he was really worried about his cash flow. So that’s why we created Fluid. Now, we position ourselves as an AI-powered payment platform with financing. We enable businesses to get paid faster, pay seamlessly, and access financing instantly. Our vision is actually to build a team of AI agents that will automate processes that currently take finance teams hours or weeks to do, to really complete them in minutes.

Paulo: And make the lives of folks like John, for instance, the seafood distribution business, a lot better. They can run a healthy business and also enjoy life, so to speak.

Use cases for AI agents in the B2B payments flow

Paulo: I wanted to touch on that goal: being able to create a team of AI agents to automate specific processes that previously finance teams had to spend so much time on. Maybe we can talk about what those specific processes are.

Trasy: Ah, okay. Let’s use our AI reconciliation agent as an example. Currently, the finance team has to manually go to Excel and tally those transactions one by one. You can think about running a distribution business—you have thousands, tens of thousands, or 50,000 invoices a month, and you have a team of finance people just tallying the invoices in Excel. After they’ve done it, they will have to upload that back into the accounting system, or in some cases, they have to key it one by one in the accounting system.

How our agent does it is that it uses logic and reasoning to calculate and really understand which invoice the customer has actually paid. The agent keys in the payment and automatically tallies it, like adjusting the outstanding balance. So the finance team doesn’t really have to do it. What takes the finance team seven to eight hours a day to do, now with the agent’s help, they are superhuman and finish it within 30 minutes.

Our team also benefits from that because we have a lot of invoices as well. With the AI agent’s help, we can really finish our job with like 20% of the effort. And that’s also why we are able to actually scale our business without really scaling the team, thanks to the AI technology.

Paulo: It is really great if you yourself are able to use the kind of solutions that you’re building and really speak to how effective they are. Are there any other use cases that you’re currently launching or in progress?

Trasy: Yeah, we have been using the AI agents for months already. Now we are in limited testing with some of our distributors and launching it to get initial feedback from their perspective. Of course, for us, we use just one accounting system, but for them, there are different types of accounting systems.

I’m really excited to see that they are very open to adopting new technology in such a traditional industry. Of course, we have a number of other agents in the pipeline. The other one is the AI collection agent. Right now, Fluid and the distributor have to pretty much collect manually and chase payment via WhatsApp, email, or phone call. What we are planning to build is really an AI agent to do collections specifically for the industry with a lot of invoices. The agent can send reminders and negotiate a payment plan for the B2B businesses in WhatsApp, email, and potentially phone calls.

Actually, it’s interesting, I just talked to a startup here today that will be able to identify whether you have the willingness to pay back or you have the money to pay back with just a voice. That’s super powerful, and the accuracy is over 93%. So I think what we are seeing is a lot of distributors who have outstanding receivables will be able to actually plug in our agents and use it for their own benefit.

Paulo: Managing accounts receivables can be a really daunting task, and the impact can accumulate, especially with a lot of different AR invoices. So it’s going to be useful.

Learnings from driving AI agent adoption for SMEs so far

Paulo: Any particular learning so far from talking to a lot of these enterprises? It seems like you really get a lot of feedback from them and have these conversations. Anything that you learned over the past year?

Trasy: The AI agents are definitely something that they all have been thinking about. When we were first thinking about this in a traditional industry with new technology, we were very cautious: “Are they actually ready?”

But a lot of the older owners of distribution businesses have actually been using AI in the last one year. So they have already been thinking about it.

Paulo: So the personal use case has actually been helpful in adoption?

Trasy: It’s like my mom, honestly, using AI in China, and it’s a personal context. So even parents at this age are adopting AI technology very well. For business owners, even in the more traditional companies, they have been thinking about how to improve efficiency for the business, how to collect more money, and how to improve the collection of accounts receivables. So I think that has been very encouraging.

Biggest challenge so far driving AI agent adoption for SMEs

Paulo: What was the biggest challenge on the other hand so far? It seems like they’re ready to adopt, but are there any challenges in actually executing or implementing that?

Trasy: Yeah, definitely. I think one of the challenges would be how it works for others. It’s a five-minute connect to a cloud accounting system.

Paulo: What types of accounting systems do you guys work with?

Trasy: It’s like Xero, QuickBooks, Microsoft Dynamics, SAP, and Oracle. We are actually connected to all of these. It’s just one quick connection.

Paulo: Flexible, a lot of options.

Trasy: Very. I would say it’s a lot easier. But then the challenge is that in some older companies, they don’t really have a cloud accounting system. So I think it’s a question of, should we actually go that route or should we actually just stick with the cloud-based systems? What we realize, especially in emerging markets, is that there are still a lot of legacy systems. And that is making the scale a little bit challenging.

So we need to figure out if we should really spend time building the middleware, or should we really just focus on the cloud users.

Paulo: Because that can take a while also, to transform the on-premise guys into cloud users as well. That’s a whole other transformation in and of itself.

Trasy: Yeah, exactly. I think that would definitely take some time. And that’s also the same for our credit agent, which is basically the financing side of Fluid. When we connect to the accounting system, the top-tier ones are very quick, like five minutes, but the non-cloud ones definitely take a few days, sometimes like weeks.

Competitive advantage for Fluid with its development of an AI agent stack

Paulo: And for Fluid, how do you see this move into AI agents and building this stack of agents—I guess that’s the longer view—that finance teams can rely on? How does that reposition Fluid within the competitive landscape compared to others in the space?

Trasy: Yeah, that’s a great question. I think for us, our vision never changes, which is to make financing less fluid. We really want to build the biggest indicator network so that businesses can get paid faster. We want to be the best possible domestic cash flow partner. So instead of really thinking about the more traditional way of how to make it happen, with AI, we are able to actually use AI to think about how to make it happen. When we see the success of a number of AI agents, we realize that’s probably the way to go.

So that’s how we are going to differentiate ourselves from others. Even when you talk about AR/AP management, we really need to think about how to make it so interactive and easy for our users to use and adopt. And at the same time, if they need financing, they can also stay positive and get paid on day one, which they could do now. We are interacting in a more interactive way and an easy way for them to adopt. I think that will be very sticky.

Paulo: So five minutes to onboard, and then it runs in the background already. And as you mentioned earlier, tasks that take days, even weeks, if you’re dealing with a lot of paperwork, can take 30 minutes. That’s a really great difference.