Latest / The Payments Shed / Ep. 62 - LIVE from FTT Payments with Payrails: The Payments Execution Layer - Why Infrastructure Is Replacing Platforms

Transcript

Ep. 62 - LIVE from FTT Payments with Payrails: The Payments Execution Layer - Why Infrastructure Is Replacing Platforms · The Payments Shed

• 0:04
  Welcome to the payment Shed Podcast live.
• 0:07
  Thank you to those of you who have stuck around till the end
• 0:09
  of the day. We very much appreciate it.
• 0:11
  For those of you who don't know us, we are a payments, fintech
• 0:15
  and business leadership podcast streaming on all of your
• 0:18
  favorite platforms. And today we're fortunate to be
• 0:21
  joined by Edward Moore, CRO at Pay Rails.
• 0:25
  Edward, please introduce yourself to us a bit about your,
• 0:27
  your background and your role at Pay Rails.
• 0:29
  Call me Ed. My mum used to call me Edward
• 0:31
  so. It's just the formal so people
• 0:33
  can find you on LinkedIn, you know?
• 0:35
  So Ed Moore, I'm the CRO at Pay Rails and so.
• 0:39
  What is pay Rails? Pay Rails.
• 0:40
  Is basically a modular payments infrastructure platform and so
• 0:44
  the whole thesis of the company. Was to decouple the payments.
• 0:49
  And and to enable merchants or enterprise merchants to have
• 0:52
  control in this context of what decisions they make, but also to
• 0:56
  help them scale globally and acceleration, simplify the
• 0:59
  access to payments regardless of what country, geography, payment
• 1:02
  method over that element that you want to work with.
• 1:05
  Fantastic. So maybe to start it all off,
• 1:09
  there's a lot of talk about this idea of the payments execution
• 1:13
  layer. Maybe you can give us a little
• 1:15
  bit of an insight into what that actually means.
• 1:18
  Yeah. So what the payments execution?
• 1:19
  Layer is. It's basically the.
• 1:20
  Doing parts of payments. So it's responsible for moving a
• 1:24
  transaction from intense to outcome.
• 1:26
  And so if you sort of dig down and sort.
• 1:28
  Of the categories, there's basically.
• 1:30
  The transaction life cycle. So like authorization, capture,
• 1:34
  refund? Voids chargeback.
• 1:36
  Then you can look at like provider abstraction and so
• 1:39
  obviously when from our contacts there might be multiple.
• 1:42
  Providers or PSPS that you work with.
• 1:44
  And so defining that workflow and having that ubiquitous
• 1:47
  across. Regardless of the.
• 1:48
  Provider and that flow. Then there's tokenization,
• 1:51
  obviously how you want to. Capture those card details.
• 1:53
  And present. Those to their respective PSPS
• 1:56
  or acquirers. Then there's 3DS, then there's
• 1:59
  SCA, then there's failure handling.
• 2:02
  So if a transaction fails, how do you want to handle that?
• 2:05
  Do you want to? Do a retry.
• 2:07
  How do you want to retry that? How do you want that flow to go?
• 2:09
  Then there's settlement and then there's reconciliation.
• 2:13
  And do you think that the next decade sort of belongs to
• 2:17
  modular infrastructure sat beneath software?
• 2:19
  Is that where your mindsets go in with this this product?
• 2:22
  Yeah, look, I think for. Pay rails like our prime.
• 2:25
  ICP is generally enterprise. Customers or customers who are
• 2:28
  global by. Design or have a massive
• 2:31
  expansion context and so. To do that.
• 2:34
  There's no one PSP or provider in the world that can stitch all
• 2:38
  of that together. And so that's where this.
• 2:40
  Advent of infrastructure layers came, but primarily we work with
• 2:45
  lots of different customers and depending on the maturity and
• 2:48
  the investments they've made in their technology stack, you
• 2:51
  know. I worked at Stripe prior.
• 2:53
  To being at pay rails and the common narrative was depreciate
• 2:56
  everything you've invested in and move everything over to us.
• 2:59
  And for a lot of merchants, that was a fundamental problem and so
• 3:02
  they wanted to. Decouple the.
• 3:03
  Experience I want to pick the best and breed parts that I
• 3:06
  either don't have or I don't want to maintain or I'm not good
• 3:09
  at and I want. To use third parties to.
• 3:11
  Extend that out. And so I've seen that for a
• 3:14
  couple of years since I've been in payments and sort of that
• 3:17
  notion of decoupling and using different components to
• 3:20
  complement your technology stack is what enterprises are looking
• 3:23
  for. Absolutely so.
• 3:25
  But now especially you see so many different merchants have
• 3:29
  been sold the story of course like you mentioned already at
• 3:31
  Stripe you, you have one integration, you get all of
• 3:34
  these financial infrastructure products all together.
• 3:37
  You have a couple of other companies who are all who are
• 3:39
  all providing that same stack. Are you seeing merchants
• 3:43
  actually pushing back on that all in one provider and trying
• 3:46
  to look much more towards a modular type of stack so that
• 3:49
  they can decide which way and which direction they can go?
• 3:53
  Yeah. Look, I think, look, I think
• 3:54
  fundamentally the answer is there's room for both.
• 3:56
  But I think depending on the nature of a business and the
• 3:59
  scale, look, I think when you're with an all in one provider,
• 4:02
  there are compromises, There might be compromises and
• 4:05
  authorization performance. There's definitely compromises
• 4:07
  in price. That's a big.
• 4:09
  Part, but also geographical coverage.
• 4:11
  And all of performance is generally, I suppose the
• 4:15
  downward impacts. If you go with an all in one
• 4:17
  provider, there are benefits. And thus it simplifies
• 4:20
  reconciliation, it simplifies payment operations, but there
• 4:23
  are pluses and -2 thus, and depending on the nature of the
• 4:26
  business, then that compels them to say, hey, I, you know the
• 4:30
  general thing. I I.
• 4:31
  Hear a loss is I'd love to have a stripe that's global, but
• 4:34
  it's. Cheaper and it's performative
• 4:35
  across the board. Sure, great.
• 4:37
  I think everyone fundamentally identifies that gap.
• 4:40
  But nobody does this. And So what we're intending to
• 4:42
  do is to enable a merchant depending on the maturity of
• 4:46
  their payments, their payment stock is to sustain what they
• 4:49
  want to keep and then use different components with
• 4:51
  ourselves. Which part of that do you think
• 4:54
  is the the thing that tends to break first though?
• 4:57
  Yeah. Look, the major, major reason
• 4:59
  for looking at alternative like payrolls is authorization
• 5:02
  performance. If you think of like AI, which
• 5:05
  is a great example, every AI company by design is global from
• 5:08
  day one. And so they need to be in every
• 5:10
  single market, in every single jurisdiction.
• 5:12
  And you know, I was in San Francisco last week speaking to
• 5:15
  a number of those companies, and the moment they scale, they've
• 5:18
  got problems all over the place. Authorization performance
• 5:21
  conversion is a big, big element for them.
• 5:24
  And so if they're not capturing share, they're losing.
• 5:26
  And so they're in a hyper growth mode.
• 5:29
  And so they're thinking of like how do I do I have conversion at
• 5:32
  checkout? How do I capture a customer when
• 5:33
  they want to buy? And then there's lots of
• 5:36
  ancillary downstream problems. Now they're running into is
• 5:38
  like, how do I, my costs are going through the roof?
• 5:42
  How do I manage those and tax and compliance, how do I manage
• 5:45
  those factors? So on and so on and so on.
• 5:47
  And so there's this whole myriad of problems because they're
• 5:50
  scaling globally. And again, if you go back to the
• 5:52
  point of there's a certain point in time for a lot of these
• 5:54
  companies where they're scaling and they're in hyper growth
• 5:58
  mode. But now they're starting to see
• 5:59
  all the problems coming out and they're not looking for
• 6:01
  solutions, but the problem for them is in a lot of cases they
• 6:05
  don't have payment capability and so they're looking for
• 6:08
  advice. What do I do?
• 6:09
  I don't know where to go. And and then unfortunately,
• 6:11
  because fragmentation and payments, depending on who you
• 6:13
  speak to, you get many, many different answers because we're
• 6:16
  all coming from a different position of well, I'll actually
• 6:19
  everyone's intending to help, but depending on where you're
• 6:22
  at, you'll get many different answers.
• 6:23
  I think the authorization piece is really interesting though,
• 6:25
  because we are seeing more merchants and platforms move to
• 6:29
  a multi supplier setup, right, Particularly from the
• 6:32
  globalization point that you make.
• 6:34
  And invariably certain acquirers, for example, have
• 6:37
  better authorization rates than than others.
• 6:39
  So you might have to move into a new market with a different
• 6:41
  acquirer. It's the one you're used to,
• 6:42
  whether that be described or whoever that, that you
• 6:44
  reference. But you're then probably going
• 6:46
  to see a material impact against what you're used to.
• 6:49
  And then businesses have to work with brands to try and increase
• 6:52
  that number to more familiar territory, right?
• 6:54
  So how? How do you tend to present that
• 6:55
  authorization data back to your customers when you're working
• 6:58
  with them? Yeah, look, so.
• 7:01
  If if a merchant is working with multiple PS, PS, we generally
• 7:04
  would ask them for the settlement files and do an
• 7:06
  analysis on their general decline reasons and their setup.
• 7:10
  I think fundamentally you look at obviously payment operations
• 7:14
  and then you end up going to the office of the CFO saying how do
• 7:17
  you handle reconciliation? How do you manage manage this
• 7:20
  process end to end? And invariably it's broken
• 7:23
  across the board in a lot of places.
• 7:25
  So the common narrative is you see outsourcing into lower cost
• 7:28
  markets where people are trying to get an army of.
• 7:31
  Lower cost resources. To manage these payment
• 7:33
  operations fundamentally our position or.
• 7:36
  You know where? We go in to support merchants is
• 7:39
  number one we. Look at if you use.
• 7:40
  Alternative providers in different jurisdictions, what
• 7:43
  authorization uplift can you see based on your decline reasons?
• 7:46
  That's one, the cost piece we don't control, but obviously
• 7:49
  intuitively you know, depending on the markets that you move
• 7:52
  into and the scale of the merchant you're in, you can you
• 7:55
  can get better economics, but that's not something we can
• 7:58
  obviously guide them to that point, but we don't negotiate
• 8:00
  the contracts. They need to figure that out
• 8:02
  themselves. The big piece we focus.
• 8:04
  On is that payments execution layer?
• 8:06
  What are the attributes that we can solve where we can create a
• 8:09
  single console that you can control things like 3DS
• 8:12
  agnostically, you can control your token strategy, you can you
• 8:16
  have. Clarity about what your retry?
• 8:18
  Strategy is and that differs market by markers because in
• 8:20
  different markets there are different retry attributes you
• 8:23
  want to do. So for example, we have a
• 8:26
  customer in Saudi Arabia where we were retrying instantly that
• 8:31
  the payment would fail and we try again and we go back to the
• 8:33
  same account. But what we found is a lot of
• 8:35
  people get paid daily and they get paid after 5 O clock.
• 8:38
  So we said OK. We'll retry everyone five O 5
• 8:41
  and guess what, you know our our success rate went up.
• 8:43
  Through the roof because. People have money in their bank
• 8:45
  accounts. So it's those.
• 8:47
  Little nuances that you need at a country by country level.
• 8:50
  Yes, you need to understand payment operations, but also you
• 8:53
  need to understand the heuristics of a Pacific market.
• 8:57
  Absolutely. I want to dive a little bit
• 8:58
  deeper on orchestration especially, right, and
• 9:01
  orchestration has been a big team for definitely over the
• 9:04
  last five years. Why isn't that enough anymore?
• 9:09
  Yeah, look, I think like. There is a big opportunity.
• 9:12
  For orchestration in the Marcus and so obviously we're pursuing
• 9:15
  it like. Other very well heeled
• 9:17
  competitors. And but I think that the
• 9:20
  predominant. Piece is is there's sort of
• 9:21
  three. Pathways you can go with
• 9:23
  orchestration you can do nothing and suck it up and take the
• 9:27
  pain. And which a lot of merchants,
• 9:29
  actually. Do to you can build it yourself
• 9:32
  which you know the, let's call it the.
• 9:34
  The mega enterprises, you know, have a preference of control.
• 9:38
  So they, you know, I remember Porsche have built their own
• 9:42
  orchestration platform and the the number one reason is they
• 9:44
  wanted to control everything. They didn't want any third party
• 9:47
  involved. And so for certain companies
• 9:49
  they don't want a dependency on anyone.
• 9:51
  They will build it themselves and then for everyone else.
• 9:54
  There's third parties like ourselves.
• 9:55
  And and other competitors in that market and and so with
• 9:59
  infrastructure ultimately the pay off is I get improved
• 10:03
  performance that's the majority of part.
• 10:05
  And so I think that's great and but the.
• 10:08
  The thing is is if you actually go.
• 10:10
  Back to that payments execution. Layer.
• 10:11
  There are still other. Problems that aren't being
• 10:13
  solved. So while you might solve the
• 10:15
  payments acceptance piece, you're not solving how.
• 10:18
  Do you deal with chargebacks? Agnostically.
• 10:20
  How do you deal with? Reconciliation.
• 10:22
  And so there's more downstream problems that come as a.
• 10:24
  Result of that so while. Orchestration fixes 1 component
• 10:27
  of the, let's call it the payments.
• 10:29
  Execution layer it. Doesn't fix all the other
• 10:31
  components. And what happens when you start
• 10:34
  decoupling things like routing, chargebacks, tokenization?
• 10:38
  And our decoupling is obviously a big focus for pay rails, but
• 10:41
  what is the kind of long tail effects of that process?
• 10:45
  Yeah, look, I think the. Majority of part is is.
• 10:48
  For the merchants that want to do that, they want control.
• 10:51
  Or they're happy with the. Existing technology investments
• 10:54
  they have. So you know we.
• 10:56
  Become sticky in the context of we're not trying to pull
• 11:00
  everything apart. Our ultimate mission is to give
• 11:03
  control back to the merchant. Because if you actually.
• 11:05
  Look at the general fintech model today.
• 11:08
  It's a closed network model which is a traditional SAS model
• 11:10
  which is unlock the customer in and make it as difficult for you
• 11:14
  to move somewhere. Else our core.
• 11:16
  Principle is, is we'll make it open, we'll make it modular
• 11:19
  where you can use the different components that complement your
• 11:21
  technology stack. And then of course, we need to
• 11:25
  drive performance. If we're not driving
• 11:27
  performance, you'll be dissatisfied and you'll move
• 11:29
  elsewhere. But we'll make it, you know,
• 11:30
  we'll make it easy. If you want to move, that's OK,
• 11:33
  right? And so that principle means we
• 11:35
  don't just need to provide the technology we need to.
• 11:37
  Provide the service because. Some of our merchants are very
• 11:40
  savvy payment professionals and others are still learning us.
• 11:43
  And so it's different levels of service levels that we need to
• 11:46
  give to those types of customers, but like the major.
• 11:48
  Part is, is that you? Can get perform.
• 11:50
  You can see performative uplift in decoupling and complementing
• 11:54
  your technology stock. I worked for Microsoft for. 15
• 11:57
  years IT was. Built on that premise of
• 11:59
  decoupling and using components to complement your technology
• 12:02
  stack that hasn't really landed into payments because of the
• 12:05
  fragmentation of the ecosystem. And by design, everyone's trying
• 12:08
  to replicate SAS, which means it's a closed network.
• 12:11
  I'll make your life very difficult to move anywhere else.
• 12:14
  And as you're using that approach, looking into the
• 12:16
  market, seeing some of the deals that you're winning right now,
• 12:19
  what are some of the the the new markets that you're or the new
• 12:22
  opportunities that you're unlocking as you're decoupling
• 12:25
  and providing that as a stand alone?
• 12:27
  Yeah, so like it would be. Fair to say we're looking as.
• 12:31
  All sort of. Fintech business models and
• 12:34
  opportunities and seeing where we can decouple.
• 12:36
  So quite recently, we've launched agnostic chargeback
• 12:39
  management. And so we're doing it from a
• 12:42
  context of consolidating the data layer.
• 12:45
  Because if you if. You control the OR control if
• 12:48
  you support the merchants and driving your execution layer.
• 12:51
  So if you are doing orchestration or tokenization
• 12:54
  and then you were building. A chargebacks product that
• 12:56
  pulls. In that data and what?
• 12:58
  We're doing is obviously we're. Using, you know, the modern LLMS
• 13:02
  and automating that process. There's two parts to US.
• 13:05
  One, there's a productivity. Gain.
• 13:06
  We see a lot of merchants who have high chargeback volume
• 13:09
  don't respond to all. Of us because it's a.
• 13:11
  Very manual process of people filling in the.
• 13:15
  I always get the word represent. Representment files and then
• 13:20
  basically issuing that off. That's a manual task.
• 13:22
  And so effectively what we've seen with merchants is, is they
• 13:26
  want somewhere that pulls in, pulls in all of those different
• 13:29
  data components that are a core decision making.
• 13:33
  Criteria of whether do you? Defend or do you just accept?
• 13:35
  Or do you move on and so? Our chargeback product is
• 13:38
  really. Around around those principles.
• 13:40
  Of it's a productivity tool. So we're not taking 30% of the
• 13:44
  win rate. We're saying, hey, we'll charge
• 13:46
  you a SAS fee will charge like A1 euro €2.
• 13:49
  Type of fee. And the reason why we're doing
• 13:51
  this is to said primarily it's a productivity.
• 13:52
  Tool, we're going to drive. Efficiencies by by consolidating
• 13:55
  your data layer of your transactions, your tokenization
• 13:59
  and then connecting with your order management data where we
• 14:01
  can automate that process and allow you to respond to the
• 14:04
  volume of. Chargebacks that you have today.
• 14:06
  So when you. Think of that like is.
• 14:08
  It infused with AI it. Is an AI tool.
• 14:10
  But what we're doing is consolidating the, the, the, the
• 14:14
  execution layered payment data that you need to inform the
• 14:17
  decision making of whether you know, is this a loyal customer?
• 14:21
  Have they been with us for 10 years?
• 14:22
  If they're just if there is a chargeback.
• 14:24
  Notification or they? Do issue a chargeback.
• 14:26
  You're probably just better off not defending it and giving it
• 14:28
  back to them. Plus you know if you bring in
• 14:31
  that order data you can validate that with AI and go Yep.
• 14:34
  This is a legitimate order. You should.
• 14:36
  Refund them the money or not, you should defend.
• 14:39
  Just very briefly on that point you mentioned sort of the, what
• 14:43
  I think is the traditional way of how we, how we treat payments
• 14:47
  and pricing especially, right. We, we talk about basis points,
• 14:49
  we talk about transactional fees, but especially within AI
• 14:53
  you now see that it's much more event based, token based.
• 14:57
  Do you see that happening in our industry as well where we, we
• 15:00
  will start looking at these processes much more of a how is
• 15:03
  the LLM, how much token input has been going in there?
• 15:06
  What's the efficiency within that and not just a take rate or
• 15:09
  a margin that's being applied? Yeah, look, I think there's a
• 15:12
  load. Of companies, you know, sort of
• 15:14
  dipping their toe and building like a payments agent to replace
• 15:18
  the payments manager, if you want to call it that, or for
• 15:20
  payments operations. I think it's more nuanced than
• 15:23
  that. I think you obviously have to
• 15:24
  build AI agents for specific use cases.
• 15:27
  I think over time, once you can prove demonstrably it can
• 15:31
  intellectually. Replace or partially replace a.
• 15:33
  Human, then yes you can, but ultimately that cost can't.
• 15:36
  Exceed what it costs to have a. Person.
• 15:38
  So I think it's it's sort of binary in a sense and.
• 15:40
  That it's fixed to a degree. There's definitely an
• 15:43
  opportunity there, but I think in terms of like.
• 15:46
  AI as a service there's. Definitely an opportunity for
• 15:49
  us, but I'm not sure. The scale of it, yes.
• 15:51
  But I think what we're focusing on is again that execution layer
• 15:54
  piece around how can we automate that and make that efficient for
• 15:58
  merchants? So so.
• 15:59
  They can be effective in what they do as opposed to harming an
• 16:02
  army to build that out. But we're very conscious of the
• 16:04
  fact that the the. Alternative to that is.
• 16:08
  You can just hire 10 people in the Philippines and maybe that
• 16:10
  will cover it, maybe it won't, but we're seeing for large
• 16:13
  merchants it doesn't. It's.
• 16:15
  They're basically just. Focusing on the 8020 and leaving
• 16:17
  the rest, what we want to do is empower them that they can
• 16:20
  defend everything that should be defendable and then or you know,
• 16:23
  with notifications, they can say no.
• 16:25
  Let's just, let's just. Refund it and let's just move
• 16:27
  on. And so we're trying to empower
• 16:29
  them again. It's empowering them with choice
• 16:32
  and productivity as opposed to saying, hey, we're going to take
• 16:35
  4050 basis points of every single thing we're doing.
• 16:38
  I think at a certain point in time, that model will probably
• 16:42
  be challenged. It's not.
• 16:43
  Yes, but at a certain. Point in time, I think it will.
• 16:46
  On on the notion of enterprise customers, do you feel like
• 16:50
  they're generally building payments teams more like
• 16:52
  infrastructure teams in 2026? Yeah, I was in the US last week
• 16:57
  and pretty much in the. US you see a lot.
• 17:01
  Of dominance with engineering managers running payments as
• 17:03
  opposed to payments teams and you are starting to see that in
• 17:06
  Europe as well. I think one thing that's really
• 17:09
  interesting is when you actually look at the cost of payments
• 17:11
  generally, I always hear the processing costs or
• 17:14
  infrastructure costs, but you never hear the people costs
• 17:16
  saying, hey, well, you've got 10 people over there managing
• 17:18
  integrations, what about them? That's also a cost.
• 17:21
  And they're like, Oh yeah, but we already have them.
• 17:22
  I'm like, no, I get it. But you don't need those 10
• 17:25
  people. You could redeploy them and
• 17:27
  build your product or do something else.
• 17:28
  With those resources. And like in a way and in a way
• 17:33
  like payment engineer engineers, there are certain engineers who
• 17:36
  want to be specialized in payments, but there's a loss you
• 17:38
  don't, they want to do, They want to build product, they want
• 17:42
  to ship product, right. And so you do see that infusion
• 17:45
  A. Lot, but you do see.
• 17:46
  A lot more engineering, particularly in the US and I'm
• 17:48
  increasingly you're start to see that more and more here where
• 17:51
  ultimately the decision makers are now becoming the engineering
• 17:54
  leads aligned with the the payments PMS. It's still the
• 17:58
  payments PMS in Europe make the decisions, but in the US
• 18:00
  increasingly it's the the payments engineers are the ones
• 18:03
  who are actually making the decisions.
• 18:05
  And because they don't want to take on technical desks, they
• 18:07
  want to be clear that it's easy and smooth operationally for
• 18:10
  them to run. Absolutely.
• 18:12
  So whenever I talk to merchants, I think one of the things that
• 18:15
  always comes back to me is that how much you need that control
• 18:19
  that they want to have. What what does it really mean
• 18:24
  for a merchant to have control and what does that look like in
• 18:27
  an open stack versus like a closed stack?
• 18:30
  Yeah, look, control is ultimately optionality where
• 18:36
  like one of the common ones I want to sell in these fifty
• 18:38
  countries, my PSP is not there. How do I, how do I go into those
• 18:41
  markets so? It's the choice of DOS.
• 18:44
  I think the other part is, is economically, I hear a lot of
• 18:48
  frustration with certain all in ones where saying people want
• 18:51
  better pricing and that they're not going to give us.
• 18:53
  And the reason why they're not going to give us is because
• 18:54
  they're an all in one solution. They say, hey, we make all of
• 18:57
  these things easy. I think a lot of merchants
• 19:00
  respectfully miss underestimates the complexity they take on when
• 19:06
  they start working with multiple PSPS.
• 19:08
  They go. This should be easy, no?
• 19:09
  Problem I'll hire one person, we'll make it all work and away
• 19:12
  we go but then what they forget about finance, they forget about
• 19:15
  all the operational components and then they start seeing an
• 19:18
  awful lot of challenges and then they start looking for somebody
• 19:20
  to help them and but I think fundamentally.
• 19:23
  Choices is about if I take the ubiquitous piece choices about
• 19:27
  the user. Experience that you want to have
• 19:29
  with your customer. You want the best buying
• 19:31
  experience. And so if you look at.
• 19:32
  Retailers. A lot of retailers.
• 19:34
  Are modernizing their buying experience.
• 19:36
  The best example of that was during COVID.
• 19:38
  You know, I personally don't carry cash anymore.
• 19:40
  I don't even carry cards when I travel.
• 19:42
  I put cards in my bag. I don't even know what the PIN
• 19:44
  numbers are. I use Apple Pay for everything.
• 19:47
  And if I go to buy something online that doesn't have Apple
• 19:49
  Pay and they ask for a credit card, I'm like, jeez, I'll just
• 19:52
  go book somebody else, right? So you can all giggle because I
• 19:55
  think pretty much everyone in this room.
• 19:56
  Probably does that. You know my kids.
• 19:59
  You know I have to pay. I had a builder in the house
• 20:02
  last week and I paid them cash and we all know why I paid them
• 20:05
  cash. And my kids are like.
• 20:07
  Jeez, I've never seen that much money in my life.
• 20:08
  It was like a couple of £100 and they're like, I've never seen
• 20:10
  that much money. That and I was like, yeah.
• 20:12
  That's that's what. Cash is and they're like.
• 20:14
  Oh, right, but like. They're not even used to it,
• 20:16
  right? So I think ultimately it's that
• 20:19
  it's the end user experience that you want.
• 20:21
  You want to delight your customer when they go to check
• 20:23
  out, when you even tend to buy. You just want to get it right.
• 20:26
  That's ultimately the the control they want.
• 20:27
  And then when you scale into markets, you want to offer the
• 20:30
  payment method that they want to use.
• 20:31
  You want to offer that preference.
• 20:33
  And then ultimately you want to get your money as quickly as
• 20:35
  possible and you want simplified payment operations.
• 20:38
  That's what everybody wants, but what everyone misunderstands.
• 20:41
  To a degree is. The complexity of scaling
• 20:44
  globally and working in multiple markets and jurisdictions.
• 20:47
  And so there's that inherent contradiction of, well, I want a
• 20:49
  cheaper, but I also want a performative, but I also want
• 20:52
  really good support and I want in every single country.
• 20:54
  What do you think this means for like merchant to PSP
• 20:56
  relationships moving forward? Because it feels like it's going
• 20:59
  from a one to one relationship to a one to one to one
• 21:02
  relationship. And that's going to be quite
• 21:04
  hard for them to manage, I think.
• 21:06
  But the, you know, the reality is that's the way the market is
• 21:09
  moving. There's a a lot of platforms
• 21:11
  that now sit in between the the traditional merchant to PSP
• 21:14
  relationship and rightly say because of the the things we're
• 21:16
  talking about today, right? Yeah, look, you know, my former
• 21:20
  life before payrolls, obviously I did embedded payments a lot.
• 21:24
  And so you, you know. Vertical.
• 21:26
  SAS platforms, you see this huge rise for SMBs.
• 21:28
  These all the ones like the flat.
• 21:30
  Pays the toast the square. Reason being is SMB is just one
• 21:33
  and all in one solution. Make it super simple in
• 21:36
  enterprises. They're not there obviously due
• 21:37
  to enterprise procurement processes that they're not going
• 21:41
  to be there by design. They're inherently making it
• 21:43
  complicated. Funnily enough, though, I have
• 21:46
  seen a certain point in time. There are merchants the same.
• 21:49
  But does that mean I have to manage the 10 PSPS?
• 21:51
  Yeah. Can you not do that for me?
• 21:53
  Well, like, well, we can, but like, no, because that's your
• 21:56
  commercial relationship with them.
• 21:58
  I think not today, but at some point in the future they may
• 22:01
  say, hey, I don't want to do this, these negotiations, I
• 22:03
  don't know what. A good deal is.
• 22:05
  Can you do it for me? I don't see that, yes.
• 22:08
  And I don't think enterprise procurement is going to let that
• 22:10
  go. But I think at a certain point
• 22:11
  in time, particularly with the evolution of AI and the context
• 22:15
  of data and pricing, and you will be able to say, well, we
• 22:18
  know what a good price is and we know what the standard market
• 22:21
  price is. I think today payment pricing is
• 22:23
  tribal knowledge. Like you guys and me, we all
• 22:26
  have an idea. Of what's a good price, but.
• 22:28
  Like fundamentally, there's a certain point in time when you
• 22:31
  will be able to see all of that data and you'll know.
• 22:34
  Exactly what people are. Paying and you can say, well
• 22:36
  look, this is these are the benchmarks.
• 22:38
  I think pricing is such an interesting topic.
• 22:39
  And we've, we've had a couple of conversations today where
• 22:42
  payments has been described as that pricing race to the bottom.
• 22:46
  And I actually disagree with with where we're at with that at
• 22:49
  the moment because I feel like we've pivoted as an industry to
• 22:51
  a point where most merchants have had a bad experience with a
• 22:55
  payments provider. And what they want is a really
• 22:58
  good payment stack that actually works for their business.
• 23:01
  And I'm seeing more and more they're willing to pay more for
• 23:03
  that. I don't know if you agree with
• 23:04
  that, but that's certainly the trend I'm seeing.
• 23:07
  Yeah, look, there's the old joke.
• 23:09
  If, if every if. Price was the major factor, we'd
• 23:12
  all be driving ladders so apologies if anyone likes a lot
• 23:14
  of not a great car apparently, but back in the olden.
• 23:17
  Days but like if price was if if cheap, cheap, cheap was.
• 23:20
  The factor everyone, we would go for cheaper everything.
• 23:23
  That's just not true. People want reliability, they
• 23:27
  want dependability, they want good service.
• 23:30
  They want people to. Even if you're a single PSP,
• 23:33
  they want the PSP. To make.
• 23:34
  Take all the complexity away that they don't have to deal
• 23:36
  with saying. It's set up, it works.
• 23:39
  Great. Thanks very much.
• 23:40
  That's all they want. And then if they want to scale
• 23:42
  and accelerate their business, they want concepts about, you
• 23:44
  know, how can I improve conversion or how can I grow in
• 23:47
  new markets. The topic about agent to
• 23:49
  commerce, it's a great, great concept.
• 23:51
  You know there's no use. Case for it yet today.
• 23:53
  But like intellectually, if it works, it becomes this new
• 23:55
  alternative Ave. to. Google to grow to drive.
• 23:58
  Growth. Cool.
• 23:59
  That's an and so it's an amazing innovation.
• 24:01
  In theory, if it works because it creates this other vector.
• 24:04
  To drive growth with consumers. Brilliant.
• 24:06
  And will people pay for it? Yeah, they will.
• 24:08
  They'll pay a premium. For us.
• 24:09
  So I do agree with you, but there is this unsavory.
• 24:13
  Notion to a degree in payments where particularly with
• 24:17
  enterprise procurement. They'll push cheap, cheap,
• 24:19
  cheap. And then when you speak to your
• 24:21
  competitors and if you lose you go, they must have gone really
• 24:24
  cheap because we didn't go that cheap.
• 24:25
  And so, you know, when I was at Stripe, we always had a
• 24:27
  perception everyone was cheaper than us.
• 24:30
  And then they had a perception we were going cheaper than them
• 24:33
  or we were taking more risk. And so I think part of that is,
• 24:36
  you know, neither I'm not in not in that world.
• 24:39
  I can see it's like you play everyone off everyone to create
• 24:42
  that perception so you can get a good price point.
• 24:45
  I think the major part for vendors is having belief in
• 24:48
  their platform and understanding the economic, the net economic
• 24:51
  gain it gives and then holding the line.
• 24:54
  And that's hard, particularly in acquiring where it can be a
• 24:57
  perception of a race to the bottom, but it just depends on
• 25:00
  the industry and the market and the value you're bringing.
• 25:02
  So. I do agree with you that.
• 25:04
  It's I sorry, I disagree with the sentiment that.
• 25:07
  Cheap, cheap, cheap. Is where it's going.
• 25:09
  Payments will be a commodity. When I was at Microsoft everyone
• 25:12
  said hosting compute would go to 0.
• 25:15
  That's complete rubbish. The margin 7580% and there's
• 25:18
  only three major players in the world, AWS, Google and
• 25:21
  Microsoft. Right?
• 25:23
  Everyone said it would be democratized.
• 25:24
  It hasn't in payments. Payments is.
• 25:27
  You could argue in person payments is a commodity, but
• 25:30
  like you still need us and you still need service providers and
• 25:33
  they still need to make money. So like the answer is, it's not
• 25:35
  0, but the answer is, is where in the spectrum is that and are
• 25:38
  you willing to participate in it or not?
• 25:41
  So maybe to come back to the whole AI part, I think we've had
• 25:46
  in the panels on agentic commerce, there's a lot of
• 25:50
  conversation. I don't want to call it hype or
• 25:52
  noise specifically, but I do believe that AI is broader than
• 25:57
  just agentic commerce within your company.
• 26:00
  Are you seeing the application of AI in machine learning or
• 26:04
  other ways of calling it that are actually bringing real
• 26:07
  returns? Yeah.
• 26:09
  So like, I suppose there's two vectors for US.
• 26:13
  One is in smart writing. So obviously we're using AI, you
• 26:17
  know, there's forms of ML and AI in mass.
• 26:20
  And so one thing I would say is, you know, when the AI race came
• 26:23
  out, everyone who was doing ML models was saying now it's an AI
• 26:26
  model, you know? Part of that was a little.
• 26:28
  Bit of spin but and that was grace by the way, there's
• 26:31
  nothing wrong with us and so like but for AI and smart
• 26:34
  writing, yes so we see anywhere from .5 to 7% uplift with smart
• 26:38
  writing so it's intelligent writing based on learnings that
• 26:41
  we've seen in the past So like it's not you know intellectually
• 26:44
  it makes perfect sense and so. But would you allow the?
• 26:47
  AI to make all of the smart writing decisions without
• 26:51
  controls. No, we still have customer
• 26:54
  success managers and payments teams making sure that it's
• 26:57
  doing what it should be doing. And so like we're not, it's not
• 27:00
  uninhibited. There's still a human in the
• 27:01
  loop today. I think over time, you know,
• 27:04
  once you get more trust with those systems, because this is
• 27:06
  payments, it is money. And so you want to make sure
• 27:10
  that you've got a strong dial on that.
• 27:11
  And so it's not uninhibited. Access where the AI is.
• 27:14
  Free rein, I think the other one is.
• 27:16
  Is is the obvious? One where you see Anthropic
• 27:18
  growing a lot, which is integrations and IT refactoring.
• 27:21
  And we like everyone else is doing that.
• 27:23
  You know, integrations, you know, we're putting, you know,
• 27:26
  can go from six. Weeks to a week.
• 27:28
  You know, once you put the AI agent on us, then it can
• 27:30
  specialize in doing those vectors, you know, the future of
• 27:33
  where it could go. You know, understanding scheme
• 27:35
  rules, You know it's a. Very deep book.
• 27:38
  You know, you guys might understand this.
• 27:39
  I'm still going through the pages.
• 27:41
  But like. If you think of like, how do you
• 27:43
  democratize? Access to this?
• 27:44
  Create a scheme, A scheme. For each of the schemes, create
• 27:47
  an agent that can say. Here's how you optimize it.
• 27:49
  For your for your use. Case for your business and the
• 27:52
  same could be on authorization rates and performance.
• 27:54
  Where's the best markets? Where should I go?
• 27:56
  What PSP should I use? What's the retry logic?
• 27:59
  That we know is most successful for my industry, blah, blah,
• 28:01
  blah, blah, blah. I think ultimately over time
• 28:04
  you'll start seeing these cases by industry, by business model,
• 28:07
  and I think all those applications make sense.
• 28:11
  Payments is a constantly moving target right?
• 28:13
  There's always new regulation coming in, there's vamp,
• 28:15
  there's, you know, God knows what every few months something
• 28:17
  news announced. So even if it can help from an
• 28:20
  education point of view is can you explain this new rule from
• 28:23
  the schemes to me and put it in common English to actually know
• 28:26
  what the Hell's going on? Right.
• 28:27
  Like, I think there's, there's definitely an opportunity there.
• 28:30
  I want to talk about the ultimate chargeback defence
• 28:33
  benefits that could exist as well.
• 28:34
  Obviously chargebacks is, is something that, you know, we
• 28:36
  talked to you guys in, in Germany about recently and I, I
• 28:39
  know that it's, it's a very passionate part of the ecosystem
• 28:43
  for pay rails. So how are you potentially
• 28:45
  utilizing AI to assist in that area?
• 28:48
  Yeah, it looks so. Alex, Alex and Vish are.
• 28:51
  Who obviously LED. Building the product, are deeply
• 28:54
  passionate about this product. Look again back to the payments
• 28:57
  execution layer. You know, we're looking to solve
• 29:00
  another component of that. There are other players in the
• 29:02
  markets. The biggest difference is
• 29:04
  they're using legacy LLMS, but also they're not consolidating
• 29:09
  the data layer that's important to inform the.
• 29:12
  AI to make the right. Decisions and so.
• 29:15
  Because we do. Orchestration because we do
• 29:17
  tokenization and. We can use those data.
• 29:20
  Sources and integrate that into the.
• 29:21
  Chargeback product as well as. Integrating into the order
• 29:23
  management system of the merchant and using AI basically
• 29:28
  aggregate those data sources and then.
• 29:30
  Teach the AI how to. Inform when we should defend,
• 29:33
  when we shouldn't? When we should initiate the the
• 29:37
  the refund. But if you think of like
• 29:39
  notifications, we're also doing that too where we have a pre
• 29:42
  notification coming into the tool, we're like going, well,
• 29:44
  hey, look, based on the history, based on everything we know,
• 29:46
  based on what we see in the data you.
• 29:48
  Should just. Refunded before it becomes a
• 29:50
  chargeback and so we're using AI Infuse.
• 29:52
  Across the whole board, we're also.
• 29:54
  Using the AI to create the representment file and so sorry
• 29:58
  I never. Get that word right?
• 30:00
  But. We're using AI basically to
• 30:01
  automate and build apps, and it's a really nice, clean
• 30:04
  document that's very succinct. That shows exactly why we're.
• 30:07
  Disputing us and what the reasons are, why we're defending
• 30:10
  and what the reasons are that it should be, giving all the
• 30:12
  evidence in a very clear format where you know the decision
• 30:16
  makes can. Look at and go yeah.
• 30:17
  OK, that makes sense. I'll, I'll along with you guys.
• 30:21
  And so again, we're trying to make it.
• 30:23
  Like very performative where? Once we connect the data.
• 30:26
  Sources. It's all aggregated.
• 30:28
  There's. The human in the loop in the
• 30:30
  context of do you want to defend or not?
• 30:34
  But ultimately then we can put assignments to the different
• 30:37
  payments operations, people that are working on on chargebacks.
• 30:41
  And so they can assign people, they can put records in it, they
• 30:44
  can categorize bits in different categories.
• 30:46
  And that whole premises is again, productivity.
• 30:48
  The main premise of AI is, you know, I like the whole notion
• 30:51
  of. AI as services it's.
• 30:53
  Effectively a service that automates that flow where you
• 30:55
  can really point your people at the decision point.
• 30:58
  As opposed to. Pulling all that documentation
• 31:00
  together and let's be honest, that is not a very.
• 31:02
  Joyous job. It's like data.
• 31:04
  Entry and pulling different sources together.
• 31:06
  It's painful. If you can remove the painful
• 31:09
  part and just go to the decisioning part, then that's an
• 31:12
  outcome that merchants want. They want everything.
• 31:14
  Aggregated they still. Want to make the decision?
• 31:16
  But you can give them all the information that empowers them
• 31:19
  to make the decision that today the humans doing, and then they
• 31:22
  can scale to more decisions. But this is the stuff we need to
• 31:25
  talk about more, right? Like we, we can spend all day
• 31:27
  talking about agentic commerce and then you have a brand like
• 31:31
  trust my group who came on the podcast a few months ago who say
• 31:34
  stop talking about travellers. The use case for I I snuck into
• 31:37
  that last part. I did hit travel as an example
• 31:39
  again, but you know, trust my group platform with 1500 travel
• 31:42
  brands on their book saying we have a massive friendly fraud
• 31:45
  issue going on in travel on the current chargeback system.
• 31:48
  And you now want to spray 10s of thousands of bookings through a
• 31:51
  gentic at us. Like let us worry about fixing
• 31:54
  the problem that exists right now in the industry before you.
• 31:56
  You hit us with that. But you take that and you make
• 31:58
  it the poster child of of something that clearly isn't
• 32:01
  ready for it, possibly as a sector, but those tangible use
• 32:04
  cases that you talk about, it might be harder for marketing
• 32:06
  teams to maybe get on a, on a flyer.
• 32:09
  But you know that is really powerful in a context where AI
• 32:13
  is is working for the better right now, right?
• 32:15
  Look, I agree, I think, you know, I think even with the AI
• 32:20
  use case of. Travel, I said.
• 32:21
  Sure, but what I'll? Do is I'll check booking.com,
• 32:23
  check the price, then I'll check the hotel, then I'll ring the
• 32:26
  hotel and say I'm booking on. I'm going to book on booking.com
• 32:29
  unless you give me a cheaper price and half the time they
• 32:31
  give you the cheaper. Price right, so.
• 32:33
  And so my whole point of that is is like AI definitely has its
• 32:37
  place and scaling over time, but even like it's not even trust of
• 32:40
  payments. It's like, am I actually getting
• 32:42
  the right price right? And like, that's a question.
• 32:44
  Mark like if you took. Everything you saw online as
• 32:47
  gospel, then fantastic for the for the merchant.
• 32:50
  But as a consumer, are you getting the right deal?
• 32:53
  I'm not so sure. So I think we got to wrap it up
• 32:59
  a little bit as we've been having this conversation.
• 33:01
  So I want to make sure that we we leave that which is the final
• 33:05
  part within the park as we have a segment which we call the
• 33:09
  Shelf of shame, which is much more focused on if there's
• 33:14
  anything in payments right now. We've talked about a genetic,
• 33:16
  we've talked about AI, we've talked about orchestration, but
• 33:19
  what is something that's been in a nuisance or an ick as they
• 33:23
  called it nowadays that that you really think that, hey, we
• 33:25
  should really believe in this behind as we move forward in
• 33:29
  payments? What would that be?
• 33:31
  Yeah, I was. Briefly talked to Grant earlier
• 33:33
  and I mentioned Recon, but I'm not going to use that one
• 33:36
  because I think AI hopefully. Will solve all of that.
• 33:38
  Over time, I think probably. One of the biggest, biggest
• 33:42
  areas that I see a. Loss is understanding your
• 33:45
  payments cost and if you work with multiple PSPS,
• 33:48
  understanding. So if you actually say.
• 33:50
  What am I actually paying and what will my?
• 33:52
  Cost be that's bloody hard to figure out.
• 33:55
  Because scheme rules different jurisdictions, payment method
• 33:58
  mix FX coming in international. Cards, corporate cards.
• 34:02
  Like it's really. Really difficult to understand
• 34:04
  this. We speak to a lot of merchants.
• 34:06
  They're they, they know what the final bill is, but they don't
• 34:09
  actually know primarily what they're paying for.
• 34:11
  So I hear that a lot. And but also when you actually
• 34:14
  look at the business case of payments, I actually spoke to a
• 34:16
  company yesterday and they said, oh, we're making money on us.
• 34:20
  I said, but how many engineers have you got on us at all?
• 34:22
  And what I was joking earlier on, they had five engineers on
• 34:25
  and I said, yeah, but you haven't factored that into the
• 34:27
  cost. So you're actually losing money.
• 34:28
  And then what's your compute costs and so on and so on.
• 34:31
  So like I think fees would. Probably be the biggest piece.
• 34:34
  It would be beautiful if you could have.
• 34:35
  A very simple descriptor of, you know.
• 34:38
  Per PSP, this is. What my actual costs are and if
• 34:40
  you could actually. Forecast that that would.
• 34:42
  Be amazing, but that's a massive problem I see from merchants
• 34:46
  today and Understanding Cost is probably my biggest book there
• 34:48
  as well. I'd like to know how you target
• 34:50
  your commercial team because you've talked about OpEx quite a
• 34:52
  bit today. I feel, I feel like their
• 34:54
  targets maybe reflect the business cost in a big way.
• 34:58
  Yeah, well, look, ultimately. We're trying to solve something
• 35:01
  and so. You have to be very.
• 35:02
  Clear about what you're solving for.
• 35:04
  And so today you know. Like I said.
• 35:07
  You know that payments execution layer notion is, is we're
• 35:10
  looking at that whole journey and saying what can we solve for
• 35:13
  for the merchant? But it has to be modules because
• 35:15
  not everybody wants every single component.
• 35:18
  Some merchants want different facets of us and and that's.
• 35:21
  The place where we're. Playing and so we compete with a
• 35:23
  lot for different players in the individual modules, but
• 35:25
  customers want optionality across those things like a
• 35:28
  common one might. Be I don't want.
• 35:30
  To do orchestration now but I wanted to tokenization I want to
• 35:33
  build that level of independence and then over time I might want
• 35:36
  to move to that sure Or you know you can use us or you can use
• 35:39
  somebody else right and so. That's up to us to.
• 35:42
  Prove that we can deliver, we can do the business, so to speak
• 35:45
  and that does come down to business cases and that does
• 35:47
  come down to ROI. But also ultimately it comes
• 35:50
  down to confidence, right? Do you have confidence we can
• 35:52
  deliver or not? I.
• 35:53
  Think your pricing point was an interesting one as well because
• 35:55
  we we've had this massive shift towards interchange plus plus
• 35:59
  pricing, which is fed down the funnel.
• 36:01
  And you can speak to small merchants that are on
• 36:03
  interchange plus plus pricing and they're told, oh, you're on
• 36:05
  the most transparent pricing structure.
• 36:08
  But all they care about is what's my effective blended rate
• 36:10
  and what are the miscellaneous charges that you're applying to
• 36:12
  my account. So I know actually what the hell
• 36:13
  I'm paying. I think that's been an
• 36:15
  interesting pivot that we've gone down funnel with
• 36:17
  interchange plus plus. Merchants have always complained
• 36:20
  about it being too complicated. And then you kind of lift it
• 36:22
  back up and go, OK, we'll, we'll move you back to blended
• 36:24
  pricing. Yeah.
• 36:26
  Now look, but I. I, I think, I'm sorry, we're
• 36:28
  probably going over, but like, I think even in a world as a
• 36:31
  consumer, there's a lot of hidden things consumers aren't
• 36:33
  aware of, right? And we know the most obvious one
• 36:36
  is DCC. You know, you go into a hotel
• 36:38
  and like, should I be paying EUR or pants?
• 36:40
  I don't which one should I be paying for?
• 36:41
  You always pick the local currency by the way, but.
• 36:43
  A lot of people get. Really, really confused and they
• 36:46
  don't necessarily understand what's going on.
• 36:49
  That is sort of like I'm using that as a point to explain
• 36:51
  broadly in payments. There's lots of
• 36:53
  misunderstandings and complexity that people don't understand.
• 36:56
  That's why like in an ideal. World you'd have.
• 36:58
  Simple transparent pricing that people guess, but obviously when
• 37:02
  you have a multi. Layered, multi party layered.
• 37:05
  Pricing model with multiple participants in US.
• 37:08
  Not everybody is as colourful. Wanting to show.
• 37:10
  Exactly what you're paying. No, agreed.
• 37:12
  We're putting on the shelf. For shame.
• 37:14
  Thank you for joining us on the the podcast today.