Latest / Leaders In Payments / Embedded Finance Special Series: Where AI Really Works in Embedded Payments with Ankita Chowdhry, Payabli | Episode 519

Transcript

Embedded Finance Special Series: Where AI Really Works in Embedded Payments with Ankita Chowdhry, Payabli | Episode 519 · Leaders In Payments

• 0:00
  SPEAKER_01: Hello, everyone, and welcome to the Leaders and Payments Podcast.
• 0:03
  SPEAKER_01: I'm your host, Greg Myers.
• 0:04
  SPEAKER_01: And as many of you know, embedded finance has quickly become one of the most important conversations in payments at FinTech.
• 0:11
  SPEAKER_01: More vertical software companies are looking beyond their core platforms and exploring how payments and other financial products can create new revenue, strengthen customer relationships, and deliver a more seamless experience.
• 0:23
  SPEAKER_01: It is also the focus of Embedify, the Embedded Finance Summit.
• 0:26
  SPEAKER_01: I'm hosting on October 13th in Lehigh, Utah, where Vertical SaaS leaders, technology providers, and industry experts will come together to discuss how embedded finance is being built and deployed in the real world.
• 0:40
  SPEAKER_01: We'll be talking not only about payments, but also lending, banking, insurance, accounting, and payroll.
• 0:45
  SPEAKER_01: You can learn more at Embedify2026.com.
• 0:48
  SPEAKER_01: As we lead up to the event, I'm talking with industry leaders who are helping software companies navigate the embedded finance opportunity.
• 0:57
  SPEAKER_01: Joining me today is Ankada Chowdry, the AI product lead at Payably.
• 1:01
  SPEAKER_01: So Ankada, thank you so much for being here and welcome to the show.
• 1:04
  SPEAKER_00: Hi, Greg.
• 1:05
  SPEAKER_00: Thanks for having me.
• 1:06
  SPEAKER_01: All right, well, let's start with a little bit about your background and how did you get into AI and machine learning and what led you to payable?
• 1:14
  SPEAKER_00: Yeah, sure.
• 1:15
  SPEAKER_00: Happy to give a bit of background there.
• 1:17
  SPEAKER_00: So I studied computer science in undergrad.
• 1:20
  SPEAKER_00: And, you know, during that time, the first couple of years are really a structured curriculum, and then you have an opportunity to take a few more elective type courses.
• 1:28
  SPEAKER_00: And I was just always really fascinated by machine learning and kind of just the way that, you know, using a ton of data and kind of mimicking how the brain sort of works with neural networks and that, you could get a machine to do pattern recognition at like very, you know, incredible speeds and things.
• 1:43
  SPEAKER_00: So I took a lot of classes at that point in machine learning, AI, deep learning, just because I wanted to satisfy that curiosity.
• 1:50
  SPEAKER_00: And then I think it was also just a little bit of luck because uh one of the internships I did uh at Facebook, they sort of match you to teams using a survey that you fill out.
• 1:59
  SPEAKER_00: And I you know wrote that I was interested in machine learning and AI, and I you know happened to luckily get matched to the ML team there building their machine learning models for ads.
• 2:08
  SPEAKER_00: So I spent the summer there sort of working on building an ML model for one of their ads things to sort of reduce the bounce rate when people are clicking through ads.
• 2:16
  SPEAKER_00: And that was my first introduction to ML in production, kind of going from learning about it in theory and then doing it in production and really enjoyed that.
• 2:23
  SPEAKER_00: So I continued by uh joining Facebook after graduating, staying on that same team, working on uh sort of capacity planning for ML models, creating the experimentation platform, and just in general, a lot of things for learning about ML at scale.
• 2:36
  SPEAKER_00: And yeah, I just continued to stay really interested in tech, but also like AI and worked at a couple of different startups.
• 2:42
  SPEAKER_00: And then ultimately I joined uh business school in in 2023, which was around the time when ChatGDPT was and kind of all the other Gen AI tools were starting to pick up a lot of steam and getting really popular.
• 2:52
  SPEAKER_00: And so I think it was a great time to be in school and have the opportunity to deep dive into those tools, you know, talk to professors and business leaders to figure out how they're thinking about implementing AI in their tools and and leveraging it and how it might change business.
• 3:06
  SPEAKER_00: And so, yeah, that was how I sort of then continued my learnings, I guess, in in the AI space.
• 3:11
  SPEAKER_00: And I knew coming out of business school that I wanted to join a company where I'd have the opportunity to do a similar kind of thing as what I did with like the ML early in the day, which is take it and understand it from school and and using it on my own time and putting it into practice at a business.
• 3:25
  SPEAKER_00: And so while I was in business school, I worked with a few venture funds and actually met an investor who ultimately is also an investor in payably, and and he's who introduced me to Joe and Will.
• 3:34
  SPEAKER_00: And so that's how I got introduced to payable.
• 3:36
  SPEAKER_00: And you know, in speaking to them early on, I think it was really awesome how bullish they were about AI and wanting to invent a future with you know AI and payments that were really deeply embedded in it.
• 3:49
  SPEAKER_00: And so it was kind of a good fit, you know, to sort of join payably and and work with them to build out their AI suite and AI tools and figure out so sort of also how internally within the company we leverage AI and and and use it to build faster and things.
• 4:01
  SPEAKER_00: And so that's kind of how I got started with AI and ML and then also ended up at Payably.
• 4:05
  SPEAKER_01: Okay, all right.
• 4:06
  SPEAKER_01: Well, thank you so much for sharing that.
• 4:07
  SPEAKER_01: So I'm really curious about and excited for this conversation because we're kind of putting together two big topics, right?
• 4:14
  SPEAKER_01: We've got AI, which everyone's talking about, and we've got the whole embedded finance area, which everyone's talking about.
• 4:19
  SPEAKER_01: So putting them together for this episode is going to be a lot of fun.
• 4:22
  SPEAKER_01: So as as you mentioned, there's a lot of talk about AI-powered payments.
• 4:27
  SPEAKER_01: So where is AI genuinely making a difference today and where is it mostly hype?
• 4:32
  SPEAKER_00: Yeah, that's a that's a fair question.
• 4:34
  SPEAKER_00: I think, you know, specifically with payments, it's important to like think about how AI, prior to all the Gen AI and sort of more like reasoning and LLMs, there was still ML, right?
• 4:44
  SPEAKER_00: And and I think that's in payments actually always been there, and it's been there for quite a while.
• 4:49
  SPEAKER_00: So in that sense, I think AI in payments and AI powered payments continues to mature in that area most prominently.
• 4:56
  SPEAKER_00: And so where it's really making a difference are in a lot of those areas where ML was already making a difference, you know, from risk, underwriting, and those types of more operational workflows that that have high volume and require a lot of interaction sort of from a human lens.
• 5:11
  SPEAKER_00: And so I think where we're seeing the reasoning models come into play is going really from those ML and rules that sort of are playing a role in in risk, and now kind of moving to how do we go from that and actually start handling items in a queue, getting a model to maybe review those before a human takes a look.
• 5:27
  SPEAKER_00: And so I think that's where we're starting to see a lot of movement and actually impact being driven from AI-powered payments.
• 5:34
  SPEAKER_00: I think in terms of hype, I would say it's more like a sliding scale where it's like more experimental versus more in production.
• 5:39
  SPEAKER_00: And I think maybe more of the experimental realm right now is what I would think is like agentic commerce.
• 5:45
  SPEAKER_00: And I think that that has been a huge topic of conversation, you know, in this year 2026.
• 5:49
  SPEAKER_00: And specifically for e-commerce, I think my take is that you know, we're seeing a lot of movement on the infrastructure side with uh, you know, new protocols or like networks coming out with toolkits for for agentic commerce and processors working with big AI labs and that kind of thing.
• 6:04
  SPEAKER_00: But I think there still is like a behavioral gap, sort of in a way, for consumers to actually start handing over their payment credentials to an agent.
• 6:11
  SPEAKER_00: AI shopping is still very much like let the AI find what I need to buy, but then I'll still ultimately come in at the at the checkout.
• 6:18
  SPEAKER_00: And so I think that the infrastructure in that realm is maybe a bit ahead of the actual volume.
• 6:24
  SPEAKER_00: But I think you know, still within agentic commerce, one area that I'm really excited about, and I think maybe we'll hopefully start seeing some more hype in in the coming months, is is agentic commerce for B2B.
• 6:34
  SPEAKER_00: Because I think like for B2B payments, like paying your vendors or you know, paying your business records and things like that, it's a much more finite solution space.
• 6:41
  SPEAKER_00: So I think like that behavioral problem is less of a hurdle to get over.
• 6:45
  SPEAKER_00: And so I think that there's a lot of interesting applications there, and I'm really excited about that particular area too.
• 6:51
  SPEAKER_01: Yeah, totally agree.
• 6:52
  SPEAKER_01: I hear a lot of people say the same thing that uh the B2B side might might lead a little bit over the consumer side.
• 6:58
  SPEAKER_01: So as we talk about embedded finance, as you well know, it creates this unique opportunity to combine both financial data and operational data.
• 7:07
  SPEAKER_01: So what becomes possible with AI when you combine those two?
• 7:11
  SPEAKER_00: Yeah, no, definitely.
• 7:12
  SPEAKER_00: I think especially like yeah, embedded finance within like vertical SaaS, you have so much operational data about like how that vertical sort of runs from end to end.
• 7:20
  SPEAKER_00: And yeah, I mean, I think it unlocks a lot of very interesting use cases.
• 7:24
  SPEAKER_00: The first one I think is is related to what I was mentioning earlier, where those workflows that historically used a lot of ML for like risk and underwriting and using that to score things, but ultimately a human still needs to go in and actually review cases that maybe fall into a gray area.
• 7:39
  SPEAKER_00: I think adding operational data that you now have and including sort of these reasoning models that we have with uh more generative AI, you can actually now, I think, get to a point where some of those cases get to be reviewed without necessarily needing human intervention at the outset.
• 7:55
  SPEAKER_00: And you can kind of gather information about those cases.
• 7:58
  SPEAKER_00: Because a lot of times I think operational data can be slightly more unstructured, maybe, than that financial data as well.
• 8:03
  SPEAKER_00: And I think you know, generative AI is uniquely positioned to be able to help make sense of that.
• 8:08
  SPEAKER_00: And so I think in some of those workflows, you you get a good, yeah, good synergy of those types of data.
• 8:14
  SPEAKER_00: But I think another way that we're starting to, at least at payably, like trying to start using both of these types of data is really in helping you know these vertical SaaS companies be more proactive in sort of understanding what they're seeing from their data and how can they take action on something in a financial sort of landscape before it becomes sort of an issue.
• 8:33
  SPEAKER_00: And so I think if a partner you know has a set of merchants that recently onboarded and you know they're lagging transactions uh or something like that compared to others in their peer set, we can use that sort of operational data that we know to sort of you know float that to the partner and kind of talk about how we can improve merchant activation or that kind of thing.
• 8:51
  SPEAKER_00: And so I think that's where you can be a little bit more unique with leveraging both types of data.
• 8:57
  SPEAKER_01: Well, as you know, in in the vertical SaaS space, you know, some of the bigger vertical SaaS companies, they they'll take on some of the risk and they'll build payments teams.
• 9:05
  SPEAKER_01: Others were will outsource that to companies like payable or other payment partners.
• 9:09
  SPEAKER_01: So, how do you see AI changing the way payments teams approach areas like risk compliance and underwriting?
• 9:17
  SPEAKER_00: Yeah, I mean I think it's in a really big way.
• 9:19
  SPEAKER_00: I think in the same way that those were the areas that got the most impacted by machine learning and and kind of helping to manage all the massive amounts of volume that come in, massive amounts of data to sift through.
• 9:32
  SPEAKER_00: I think similarly, like you know, we're seeing those areas get impacted by generative AI.
• 9:37
  SPEAKER_00: I think a lot of payments companies, you know, used to need to hire big teams to keep up with the volume that's ever growing.
• 9:43
  SPEAKER_00: But I think now with the ability for some of these tools to come in and gather additional context from those operational data that we were talking about per transaction or per case, you know, do some reasoning based on how our analyst teams were doing reasoning before, and then make a suggestion, you know, for what could happen with this case, especially in things where the you know there's lower risk, you know, starting to actually potentially auto-action on those and and uh reduce the workload, essentially, manual workload on the teams doing it.
• 10:13
  SPEAKER_00: And I think we're starting to see also from like a regulatory landscape, like more and more these tools are being, I guess, promoted to sort of have some of the you know L1 type risk things be handled more autonomously.
• 10:26
  SPEAKER_00: And so I think these are areas that we're gonna start seeing a lot more autonomous work, but it's also very important then for these tools to be built in a way that's super safe and you know has the right guardrails in place.
• 10:37
  SPEAKER_00: And it's not just sort of a tool that is spitting out anything from any LLM.
• 10:42
  SPEAKER_00: Like, you know, it's trained specifically using information on a team that has actually worked these cases.
• 10:46
  SPEAKER_00: Um and so that's also what we're thinking about at payably.
• 10:49
  SPEAKER_00: And you know, luckily we have an internal operations side that that handles a lot of the risk and underwriting for for some of our partners, and so we get a lot of learnings through that that we can then embed in agents and that sort of thing.
• 10:59
  SPEAKER_01: I grew up in the payments industry.
• 11:01
  SPEAKER_01: We always had dashboards and reports, and they tend to catch things after they've happened.
• 11:06
  SPEAKER_01: And so, you know, like how does AI help identify things or prevent them from happening?
• 11:12
  SPEAKER_01: So, what does that start to look like when AI goes from what's happened in the past to, hey, we can actually prevent it from ever happening?
• 11:19
  SPEAKER_00: I mean, I would love for that to be the case, and that's where we're we want to build towards because I think ultimately, you know, in the future, people running these payments businesses like aren't necessarily gonna have that payments background.
• 11:32
  SPEAKER_00: You know, they're gonna come in, they're building an entire software business, and payments is one leg of it, and they need to manage that business too.
• 11:38
  SPEAKER_00: And so if we can provide the ability for them to proactively know when something's gonna happen, I think that would be a huge improvement to the user experience.
• 11:45
  SPEAKER_00: And so, yeah, I mean, I think it's it's what we've been talking about really, like the combination of the operational data and the financial data and being able to have an agent that is trained to look for very specific things on like a daily basis, and then surfacing that to the customer.
• 11:59
  SPEAKER_00: You know, I think an example that you know we want to work towards that payably is, for example, disputes.
• 12:04
  SPEAKER_00: You know, if you notice like a merchant or a partner that is approaching a chargeback rate that's like slightly elevated, and then that might lead to some processing issues, you know, we could have an agent that starts monitoring those transactions, noticing when a transaction comes in that has properties that are similar to disputes that might have happened in the future or chargebacks that might have been filed in the past.
• 12:23
  SPEAKER_00: Sorry, I mean, and then you know, proactively suggesting to the merchant, you know, hey, this may result in a chargeback.
• 12:30
  SPEAKER_00: Potentially at this point, we would suggest proactively refunding, you know, and something like that.
• 12:34
  SPEAKER_00: And so you can get to a point where you can protect the merchants and the partners who might not have as much payments and finance experience to be able to do things without necessarily needing to wait until they've happened.
• 12:45
  SPEAKER_01: Well, this next one is kind of I think a big question that there's a lot of conversations around.
• 12:51
  SPEAKER_01: So for a vertical SaaS company, how should they decide what AI capabilities they should build themselves versus buy from a partner?
• 13:00
  SPEAKER_00: Yeah, that's a good question.
• 13:02
  SPEAKER_00: So I think that there's like a few simple heuristics to think about.
• 13:06
  SPEAKER_00: And I think when, you know, I think a lot of companies at this point are going to be needing to offer some sort of AI products to their customers, right?
• 13:14
  SPEAKER_00: For those same reasons we talked about, like proactively identifying things, the ability to ask questions in natural language without needing to sort of hunt around, you know, a UI and that kind of thing.
• 13:24
  SPEAKER_00: And so I think every every company is going to need to have some sort of offering in this way.
• 13:28
  SPEAKER_00: And where I think it makes sense to build is really for those types of tools that ultimately are going to directly touch your end customer, and your end customer will be interacting with on a regular basis.
• 13:40
  SPEAKER_00: Whether that is, you know, a chatbot in this example might be the best one to describe, because you'd want to control the the way that chatbot interacts, the actual responses it gives, the mechanisms that it can use, what actions it can take.
• 13:52
  SPEAKER_00: And more often than not, that that chatbot or that agent that you're making accessible to your end customer is going to need to touch all parts of your business.
• 14:00
  SPEAKER_00: So that could be payments, that could be the CRM that you've built for them, that could be the scheduling feature, all those different types of things, which on your back end might have different vendors, but to your customer should look like one seamless sort of interface, right?
• 14:12
  SPEAKER_00: And so you'd want to build one AI tool that they can interface with that you ultimately control.
• 14:18
  SPEAKER_00: And I think for that to be possible, like as a software builder and building a vertical SaaS company, it's important that you pick vendors that offer you the tools to be able to embed like their operational data that you will need for your agent into their own agent.
• 14:33
  SPEAKER_00: And similarly, I think on the on the potentially on the buy side, where things are less customer-facing and more like back office.
• 14:41
  SPEAKER_00: And so again, we we talk about you know, there's a payment side, there's a CRM, there's a scheduling, you might build many other features for your customers.
• 14:48
  SPEAKER_00: If it's more on the payment side and you are managing your own risk, let's say for your business, you know, potentially purchasing like a risk ML system, a risk rule system, or in this case like a risk agent system, which I've been talking about a little bit, right, for handling the queue and managing sort of the cases that come in.
• 15:05
  SPEAKER_00: And that made more sense to buy because at that point also what you're getting is not just the agent itself, but you're getting all the sort of knowledge that's baked in from this payments company that has worked on a bunch of different types of industries and maybe more companies in your same vertical.
• 15:20
  SPEAKER_00: So you have the benefit of that more like consortium level data rather than trying to build that for yourself with your minimal sort of focus.
• 15:27
  SPEAKER_00: And so I think what it what it really comes down to for build versus buy is like how much domain expertise is required for this particular feature and do you have it, or is it better purpose from someone else, like a vendor, and how much is it going to interface directly with your customer?
• 15:41
  SPEAKER_00: Because then that's the amount of control you would want to exert on it, is is a lot more in that way.
• 15:46
  SPEAKER_00: So I think it's important also to find vendors that can enable you to do both by providing you the tools to either pick up an agent or use an AI feature off the shelf just for your back office, or to leverage and embed an AI tool like an MCP or something to gather that information, but in your own agent that you are building.
• 16:03
  SPEAKER_01: What is actually difficult about implementing AI and where does the reality get harder than the theory?
• 16:09
  SPEAKER_00: You can talk about it specifically for payments, but I think in general, like implementing AI is just hard, right?
• 16:15
  SPEAKER_00: I think it's it's difficult, especially going from, I would say, experimentation to you know productionization and particularly when you're building the experiment or you're building the harness for what this AI tool could do, it's very easy to create like a workflow, essentially.
• 16:31
  SPEAKER_00: Like an and I kind of call these workflow vaporware, in that they are, you know, they're loops that are running, but ultimately what they're spitting out is like a generic LLMs response or something.
• 16:39
  SPEAKER_00: It doesn't really have the brain that I think is necessary to actually move the needle.
• 16:45
  SPEAKER_00: And so really what it comes down to moving from experimentation to productionization is like the accuracy, the precision, the metric moving, and the impact that this AI feature or agent is having on the workflow you're trying to improve.
• 16:58
  SPEAKER_00: And so I think the way that we try to do this at Payably is we're just really metric driven when it comes to these AI features and agents, you know, rather than sort of building something that looks like very interesting and and shows a big summary of the case, you know, to tell someone, hey, we could do this or we could do this.
• 17:15
  SPEAKER_00: It's really like, okay, after that summary, what did the analyst then do?
• 17:18
  SPEAKER_00: Did they approve it?
• 17:19
  SPEAKER_00: Did they reject it?
• 17:20
  SPEAKER_00: And did it match what the agent actually suggested?
• 17:22
  SPEAKER_00: And we want to get that number, that agreement rate, really high.
• 17:25
  SPEAKER_00: You know, I think for for us to you know deploy an agent and be comfortable with it, we we have very high bar for what that metric should be.
• 17:33
  SPEAKER_00: And doing that takes time, it takes time and it takes effort as well from the internal operations team to provide that feedback.
• 17:40
  SPEAKER_00: But ultimately, like that's that's the only way I think to move from an agent in experiment mode to an agent in production mode.
• 17:46
  SPEAKER_00: So, you know, and that's why we talked about the build versus buy thing.
• 17:49
  SPEAKER_00: Doing something like you know, building your own payment-specific agent or something requires a lot of that you know precision and accuracy and testing.
• 17:58
  SPEAKER_00: And so looking to find a vendor that provides you with something that has been tried and tested and has you know guardrails configurations all in place based on rigorous internal testing, I think is is super important.
• 18:09
  SPEAKER_01: Well, payable has been building your AI capabilities directly into the infrastructure rather than sort of treating it as a separate tool.
• 18:18
  SPEAKER_01: So why did you take that approach and what have you learned from doing it that way?
• 18:22
  SPEAKER_00: Yeah, I'll start with like the why.
• 18:24
  SPEAKER_00: I mean, I think you know, generally payable has always viewed themselves and and ourselves as like you know, the infrastructure for payments, like the AWS of payments.
• 18:32
  SPEAKER_00: You can kind of do any sort of function you need to within embedding payments within payable.
• 18:37
  SPEAKER_00: And so similarly, I think you know, we view the future of payments as having AI embedded across all parts of that.
• 18:44
  SPEAKER_00: There's just so many pieces of payments that is ripe for automation, right?
• 18:48
  SPEAKER_00: There's a lot of manual workflows, there's a lot of things with a lot of data, as we mentioned about you know earlier, there's a lot of opportunities for proactivity.
• 18:54
  SPEAKER_00: So there's just a lot of areas that AI could make a huge difference in.
• 18:58
  SPEAKER_00: And so I think because of that and that view that we want payably to continue representing like you know, the infrastructure for payments, we really want AI to be at the forefront of every single piece that you know could possibly be AIDS with with payments.
• 19:11
  SPEAKER_00: And so that's kind of the why.
• 19:13
  SPEAKER_00: And then I think in terms of like how or like why what we've done to do that is really we've taken the approach of sort of building an internal AI infrastructure layer that powers all of the different AI features and agents that we have across the spectrum within the product.
• 19:29
  SPEAKER_00: And what that does for us is it unlocks not only the ability to have multiple AI features and agents across the portfolio, but it also enables internal velocity and giving our internal developers the opportunity in from different teams and that kind of thing to be able to add AI functionality to products that you know maybe didn't have it to begin with, or to include it from the outset much more quickly and with the guardrails baked in and like, you know, making sure that PII is scrubbed out, that we have cost tracking, all of these types of things, automatically built in so that it's all doing it in one seamless sort of infrastructure.
• 20:02
  SPEAKER_00: And then similarly, I think what that enables is that every new agent or AI feature that comes up compounds on itself because we can leverage things like the agent-to-agent protocol and have them actually interact with one another so you know the insights agent can talk to the risk agent, can talk to the disputes agent, and really start to actually have a world in which agents are running the payments business.
• 20:22
  SPEAKER_00: And that's what we eventually sort of want to get to.
• 20:25
  SPEAKER_00: So by building the AI infrastructure layer in that way, rather than sort of building an AI feature here and there, we feel like we can unlock not only like the future of AI and payments, but also just internal velocity and developer efficiency in that way.
• 20:39
  SPEAKER_01: Okay.
• 20:39
  SPEAKER_01: That's a great segue into the next question.
• 20:41
  SPEAKER_01: Let's talk about the future.
• 20:42
  SPEAKER_01: So looking ahead 12 to 18 months, what do you think will change the most about AI and embedded payments?
• 20:50
  SPEAKER_01: And what do you think won't really change that much?
• 20:52
  SPEAKER_00: So I'm not a fintech veteran.
• 20:54
  SPEAKER_00: I think pavely is like my first furry into fintech, but I I think what astounded me when I when I joined and sat in on sales calls and things, you know, is is there's it's obviously a very like heavily regulated space and there's a lot of things to think about there.
• 21:07
  SPEAKER_00: And there's a lot of jargon, and you know, when you're a new company starting out and you want to embed payments into your product, but you don't have that fintech background, it can be super daunting, you know, and that's why you kind of need someone on your team who has a payments background and that sort of thing.
• 21:21
  SPEAKER_00: Maybe this is my you know, a coming from the outside world view, but I think that you know eventually we'll get to a point where embedding payments should be as simple as you know asking an agent, you know, given my operational plan and my vertical that I'm operating in, what services do I need to enable, what risk profile and policy should I enable?
• 21:38
  SPEAKER_00: And then, you know, I'm turning myself on, like I'm able to do that, you know?
• 21:41
  SPEAKER_00: And so I think it's like removing a lot of the decision making, removing a lot of the friction involved in getting started.
• 21:48
  SPEAKER_00: And then similarly, like I should be able to log on and see a single pane glass view of okay, what's going on in my payments business today?
• 21:55
  SPEAKER_00: And oh, these agents have asked me that I need human approval for these specific items.
• 21:59
  SPEAKER_00: Let me go in and do those, but let me also go and see what else the agent did in the last few days.
• 22:02
  SPEAKER_00: So I know it sounds very uh sci-fi, but I feel like there is a world in which this isn't just the case for payments.
• 22:08
  SPEAKER_00: I think it might be the case in a lot of different industries where there's um, you know, some reduction of manual work and reduction of friction in order to get things set up.
• 22:18
  SPEAKER_00: Just the barrier to build has dropped so much, you know, and so I think that'll continue to be the case for for a lot of things, especially as people now want to embed payments as part of their business at the outset.
• 22:28
  SPEAKER_00: What won't change?
• 22:29
  SPEAKER_00: Well, I mean, I think the fact that it is a heavily regulated industry, like that's I think not gonna change.
• 22:34
  SPEAKER_00: You know, I think that's important for it to be there.
• 22:36
  SPEAKER_00: We're handling money movement, so it's very important.
• 22:39
  SPEAKER_00: And so I think, you know, things like risk, underwriting, dispute management are always going to be there.
• 22:44
  SPEAKER_00: And I think they'll just continue to evolve with whatever technology comes out to sort of handle those things again in a more automated fashion.
• 22:51
  SPEAKER_00: And I think we're just gonna continue to move towards there, but the actual functions that need to happen within a payments company, I don't think should change all that much.
• 22:58
  SPEAKER_01: Well, one final question.
• 23:00
  SPEAKER_01: So if you could get every vertical SaaS company and platform, bank or payments company to understand one thing about AI and payments, what would that be?
• 23:10
  SPEAKER_00: Well, I think honestly, it it's it's really just what we've been talking about, which is that I think that AI and payments, like, well, payments is just one of the industries that's just so ripe for automation.
• 23:20
  SPEAKER_00: Like having looked at a bunch of different industries and and and worked in AI and other spaces, I just think that the amount of workflows that are currently can still manually run and and can you know continue to be very uh intensive in that way are you know the perfect kind of grounds for for this type of this automation.
• 23:37
  SPEAKER_00: And so I think it's really about you know having people first of all think that they can include payments as part of their business.
• 23:42
  SPEAKER_00: So you know, embedding payments is really should be a core part of building any business today.
• 23:47
  SPEAKER_00: It's a revenue stream that shouldn't be overlooked.
• 23:49
  SPEAKER_00: But then also I think embracing automation and embracing the fact that you know a lot of this might change, but trying to build from that with that mindset from the outset and partnering with vendors that that feel the same way and so that you can kind of future-proof yourself and take advantage of the wave.
• 24:04
  SPEAKER_00: And and yeah, I'm I'm super excited about it.
• 24:06
  SPEAKER_00: I hope other people are as well.
• 24:09
  SPEAKER_01: Yeah, I think it's it's a huge part of the conversation, not just the embedded payments, the the AI.
• 24:14
  SPEAKER_01: And I think you did a great job of kind of wrapping up what people should really be thinking about.
• 24:18
  SPEAKER_01: So thank you so much for being here today.
• 24:20
  SPEAKER_01: I know your time's very, very busy and very valuable.
• 24:23
  SPEAKER_01: So again, thank you so much for sharing all your insights and being on the show.
• 24:27
  SPEAKER_00: Thank you so much, Greg.
• 24:28
  SPEAKER_00: It was great chatting with you.
• 24:29
  SPEAKER_01: You too, and to all you listeners out there, thank you for your time as well.
• 24:33
  SPEAKER_01: And until the next story.
• 24:35
  SPEAKER_01: But before we go, we have to mention Embedify, which is coming up on October 13th.
• 24:40
  SPEAKER_01: So to learn more about the event where a lot of these AI discussions are also going to be happening, please visit Embedify2026.com.