Skip to content

Palm Internal - Product Bets Preparation - 2026-02-24

Metadata

  • Date: 2026-02-24
  • Palm Participants: Emma, Gurjit Pannu
  • Type: Internal Discussion
  • Domain Areas: Product Strategy, Forecasting, Connectivity, Reporting, Scenario Modelling, FX, Working Capital, IC Activity
  • Recording: None

Summary

Context

Internal strategic discussion between Emma and Gurjit preparing for product bets decisions. Deep exploration of Palm's positioning in an AI-first world, questioning what the company's true moat is and where to invest engineering effort.

Key Discussion Points

Strategic Questions: - What is Palm's moat when general AI tools (Claude, ChatGPT, etc.) can analyze data on canvases? - Is execution (payments, trades) commoditized or still valuable infrastructure? - Should Palm build direct bank connectivity or continue layering on TMS platforms? - How deep should we go on domain-specific intelligence (FX, working capital, bridging)?

Trust vs Execution Debate: - Trust in outcomes is foundational - no execution happens without it - But once trust is built, what's next? Is execution infrastructure our opportunity? - Connectivity issues (Discogs, Personio, Digital Realty) erode trust even for long-term forecasting - Argument: Build connectivity for trust, not (just) for execution capabilities

The "Thick Middle" - Institutional Knowledge: - Capturing organizational quirks, policies, decision logs, reasoning - Making outcomes trustworthy because they're based on "how your business actually works" - Hyper-personalization enabled by lower AI dev costs - Question: Can we deliver on this institutional knowledge layer effectively?

Forecast Trust - Time Horizon Split: - ML forecasts valuable for weeks 5-13+ (strategic decisions: investments, FX hedging) - But treasurers' magnifying glass is on the next 2 weeks (operational funding) - Are we aggressive enough on leveraging deterministic data (AP/AR, payroll) for short-term? - Need both: deterministic layer (always correct) + probabilistic layer (ML forecasts)

Execution Infrastructure Questions: - Will AI agents just use existing UIs (like voice agents clicking through systems)? - Or is there value in building payment/trade execution infrastructure? - Physical dongles, bank portals, exact amounts - when does trust cross the threshold? - Treasury Cube example: moat is connectivity + execution, not pretty interfaces

UI vs Canvas Future: - When does shift happen where enterprises work in canvases only? - Still time before that, but how do we prepare? - Reporting as storytelling (Euroports board deck example) vs just charts - Proactive experiences vs always waiting for user questions

Hyper-Customization Possibility: - SaaS model historically required standardization due to high dev costs - AI enabling lower dev costs - can we build hyper-customized treasury tools per customer? - Already doing some customization (forecasting, categorization) - Could go deeper: tailor everything to institutional knowledge, policies, what matters to that org

Potential Product Bets Discussed

Table Stakes (must do): - Advanced scenario modeling (already committed, releasing naive version) - Decision and governance intelligence (just have to do it)

Real Bets (depth decisions): - Direct bank connectivity - For trust (data quality) more than execution - Deeper IC intelligence - In-house banking, collections, payment on behalf - FX risk intelligence - Exposure visibility → decisioning → execution → measurement - Working capital intelligence - Including direct/indirect bridging, DSO/DPO, AR/AP deep dives - Stablecoins & crypto rails - Buzz in treasury world, Gurjit attending seminar with Circle/Stripe - Path to revenue intelligence - CRM integration, deal-to-cash visibility

Open Questions: - How much time to spend on each vs how deep to go? - Which bets unlock different customer segments? - What's easy to build (FX exposures) vs complex (hedge accounting)?

Strategic Tensions Identified

  1. Short-term trust vs long-term intelligence: Data quality issues (statements, connectivity) undermining trust even for strategic forecasting that's working well

  2. Build vs integrate: Own connectivity/plumbing vs layer on existing systems. Fedas exists but not positioned as core offering.

  3. UI vs agent-first: When to bet on conversational/canvas experiences vs traditional dashboards. Still need both for now.

  4. Breadth vs depth: Many valuable directions (FX, WC, bridging, IC, stablecoins) - how to prioritize and how deep to go on each?

  5. Deterministic vs probabilistic data: ML forecasts are great for long-term, but operational decisions need exact AP/AR/payroll data. Not aggressive enough on the deterministic layer?

Notable Quotes

"I'm questioning whether we need to think that we are going to need to build all the infrastructure to actually execute, or if that's going to be proven easier, because we can have agents that can use existing UIs in existing systems and just click for them." - Emma

"Until someone trusts the outcomes like hands down, hand over fist... There's no execution happening." - Gurjit

"Once we've solved trust, Oh, now all we have to do is give them a button to press. Like, that's... such a simple problem... That's literally like, okay, we need an API integration." - Emma

"The question isn't should we build connectivity because it allows us to execute, should we build connectivity because it allows us to build more trust." - Gurjit

"What is Palm's moat when general AI tools can analyze data on canvases? ... Is it the tooling we provide? Is it our agent, the internal expert on Treasury?" - Emma

"If you grabbed it into forecasting specifically... When we do our machine learning forecast, a lot of the value of that ML forecast is actually more so for weeks 5 and beyond... But if you bring it forward the four weeks... That's where I keep thinking about, like, I know ML is great, but, like, we have data at the company that's going to drive the actual, like, literal, hopefully, dollars and cents." - Gurjit

"Can we be that kind of top piece that just goes in and understands based off the context we already have from you as a Treasury and the data that you have within the company and smash all that shit together and just be able to say, look, here's a confident way around how do we think about your money" - Gurjit

"If I had an agent, [at Uber] we had outsourced to Accenture... 18 people now... can an agent do a lot of that already and faster?" - Gurjit, on Treasury ops automation

"With the cost of development going down, can we really push for hyper customized treasury tools per customer?" - Gurjit

"That's basically what they're trying to replicate in Palm... you have your chart... But beneath it is an actual narrative on why is it different... can you build these leadership decks... created in 30 seconds versus a week and a half of work" - Gurjit, on Euroports board reporting

Domain Insights

Forecasting: - Treasurers focus magnifying glass on next 2 weeks (operational), not 13 weeks out - ML valuable for strategic horizon, but short-term needs deterministic data (AP/AR, payroll) - Trust built on short-term accuracy enables long-term strategic decisions

Connectivity: - Failures with Discogs, Personio, Digital Realty undermining trust - Question: Is TMS connectivity assumption valid? Even with TMS, connections fail - Direct bank connectivity may be necessary for trust, not just execution

Reporting: - Not just charts - storytelling with narrative explaining WHY - Board decks with context aligned to what business cares about (working capital vs FX) - Self-serve for quick checks vs prepared narratives for stakeholders

Treasury Operations: - Day is not just cash management - includes bank rationalization, KYC, audit confirmations, quarterly IC settlements, hedge rolling, bank fee analysis - Many low-value ops tasks (downloading statements, confirming settlements) ripe for automation - Uber treasury ops: 18 outsourced FTEs doing manual work agents could handle

Full Transcript

Them: Hey. Good morning. Afternoon.
Me: What? Up.
Them: How's it going?
Me: Good.
Them: Yeah.
Me: Yeah. In a bit of a philosophical turmoil around, like, where everything is going. And agents are going to do everything anyway.
Them: Yeah. I spent. Yeah, I spent quite a bit of time last night just kind of trying to think, like. Just get all of the different directions of we could go. Can we do what's going to happen all into, like, a. Kind of synthesize one or two blocks of, like, this is. This is it. This is how you. How we perceive. It's just not easy. There's just so much. And then I just. You know me, I just go into different tangents to start thinking about all the macho things that are going to happen and.
Me: I'm going to send you a wall of text. Sorry. Christian was asking me, but what do I think? Is hard to figure out.
Them: Yeah. Yeah.
Me: My attempt that just like. Yeah. There's so many conflicting opinions. If you want to. I don't know, if you think it's valuable to have a quick look. But.
Them: No. Y'all take a look. You sent it to me in slack.
Me: Yeah.
Them: One. Second. One second, one second. Let's see here.
Me: I'm, like, really pushing it.
Them: Awesome.
Me: Yeah, sorry.
Them: Yeah, all done. Got it. Nice.
Me: So. This is not it. I'm not done.
Them: Yeah. Yeah, it's just. Yeah, it was just how you were thinking that. Yep.
Me: As I think we are all trying to just see beyond six months.
Them: Yeah.
Me: Because I think it's important that we try to. I know it's really difficult. Like, if we could predict that, we'd be fucking rich.
Them: Y. Eah. Y. Eah. Yeah, I think, like, question I have is on the thick middle, as you call it, or the. Not the institutionalized knowledge. Piece. Right. I think that is important because that is giving us the stickiness. It's also something that should eventually continue to build trust in the company or customer because you're saying, hey, The outcomes that you're giving me are based on the reality of how our business actually works. That. That all makes sense to me. I think, though, the part about execution being commoditized, I don't necessarily agree with that part, because I think one is execution in general is changing. In terms of what is execution, I think is still tbd. We talked about the future of stablecoin. There's one one thing, but in general, the way that you, once you've come to a decision, what you want to do with that decision, I think that's where we actually have an opportunity. If you layer on the institutional knowledge, meaning the example, you gave a notion, right? I put the note in there. I was like, well, hey, if you make an investment, you're actually going into this portal first. You're calling somebody or you're typing in a trade, and you go into this portal, and then you're making a payment. You're still doing this fragmented work, Right? And so is there an opportunity to, again, bring that into a consolidated view, that you're doing execution in the context of the business? And how they operate.
Me: Exactly.
Them: I don't think anyone's really done that. Right. And that's where we.
Me: And I think we actually aligned, and that's why this is so tricky, because it's so freaking hard to articulate. I think the same thing.
Them: Yeah.
Me: I that institutional, like the judgment, like why you should do this and why it is safe to do this with a backdrop of your company's context. And, you know, all that we have 100%. I think, I mean, that agents will, I mean, I'm questioning whether we need to think that we are going to need to build all the infrastructure to actually execute, or if that's going to be proven easier, because we can have agents that can use existing UIs in existing systems that these users might have and just click for them.
Them: Y. Eah. Yeah. Yeah. Yeah.
Me: That's just a distinction. Like, do we build out the infrastructure? For all of whatever execution is and some cases here and now because it makes sense here and now. But it's just like, interesting to see these discussions. Well, anything a human can do, an agent can do. And that's actually something my partner is building right now. So they don't even use API, so they build voice. Customer service agents, right?
Them: Yeah, yeah. Yeah.
Me: They don't even necessarily build out API integrations in the toolings for each agent, for each customer. They just use agents to go into UIs of existing systems and click. Cancel the order, do the thing, you know, so it's just an interesting development that it might.
Them: Yeah. Yeah, yeah. I think that the part that I'm wondering when this domino falls is when if you talk about execution, you mentioned it in this post around like everything has to be exact. There's no, there's no kind of like roundabout is when does the domino fall where someone, a treasurer or a team says, hey,
Me: Yeah.
Them: Actually, I do trust an agent to go ahead and make the trade and settle it and confirm that the funding is there and all that. I mean, the confirmation is not fine, but, like, going into a bank portal on my behalf with my login with my name on there. Put in a fifty million dollar payment. Right.
Me: Like your physical dongles. How does the agent do that?
Them: Yeah, yeah, yeah, yeah, yeah. All that random stuff. But. But then that's where the infrastructure question comes into place, right? Because you can still make a payment without logging into a bank portal.
Me: Y.
Them: And that's kind of. That's connectivity, for example.
Me: Eah.
Them: So that's the part that. And I don't know if I have. I think it all goes back. It keeps going back. To trust we talk about all the time is like until someone trusts the outcomes like hands down, hand over fist.
Me: They're not going to execute any.
Them: There's no execution happening, you know, And. But, like, I think. I think the way that we should think about this, though, is, like, let's. Let's hypothesize. Trust is built. Like we're working on that, right? That is something that we need to continue to do. But once the trust is built, then what? Then what? And that's where the whole kind of common around closing the loop. I think makes this interesting. And how do you close the loop? Whether it's through an agent, whether it's through an interface, whether it's through a user saying, yes, I agree.
Me: I also want to just counter with this question.
Them: Click. Fine.
Me: Is that a complex problem? Like, isn't that a luxury? Like, once we've solved trust, Oh, now all we have to do is give them a button to press. Like, that's. To me, that's such a simple problem. That's literally like, okay, we need an API integration. Okay. Like, it might take a little bit of time, but it's a simple problem. You know what I mean? It's not like complex to get it right as the whole trust, the whole thing leading up to that moment. That's the thing that is our moat. Because anyone can make an integration to like a bank that also provides payments API like, it's not hard.
Them: Yeah. Y. Eah, yeah. Yeah, yeah. Yeah.
Me: Because I think I'm still stuck a little bit in this earlier stages, as in, like. There's no point in starting working on. I'm exaggerating. Don't get me wrong. But, like, the whole, like, should we start building out bank, like, connectivity now so that we can also add payments later if. If we're not even in a place or near a place where customers would, like, base those decisions off of our. I think we are actually getting closer.
Them: I was about to say. Yeah, yeah, yeah.
Me: Positioning, for example, cash positioning, I think, definitely is one of those things.
Them: Yeah, yeah, yeah, yeah, exactly. And that's. That's the piece, right? So, like. If you grabbed it into forecasting specifically and having trust in the forecast and what's generated that. The way that. And correct me if I'm wrong, because I feel like I. I sometimes am thinking of it one way and everybody else thinking of it differently, but. When we do our machine learning forecast, a lot of the value of that ML forecast is actually more so for like weeks 5 and beyond. Because there's no, like, companies typically don't have the data. Or information or the ability to say what's going to happen seven weeks from now, eight weeks from now, because there's no invoices, there's no payroll forecast that they've received. Right. And that's where we were doing good. And we continue to do better and better, which is great because then it starts allowing for more strategic decisions, especially on longer term investments, FX hedging, all that stuff. So that's, that's where some of the decision making starts kind of being imported. And trust is important. And knowing. Hey, what, what this, what Palm is telling me for what's going to happen 13 weeks from now. Is I trusted enough to make the decision of a trade that's going to settle 13 weeks from now.
Me: Yeah.
Them: But if you bring it forward the four weeks. You know, this day, this week. That's where I keep thinking about, like, I know ML is great, but, like, we have data at the company that's going to drive the actual, like, literal, hopefully, dollars and cents that really are needed today in this account, and then on Friday on that account, and then Monday on the other account.
Me: Yeah.
Them: And that's where the execution becomes really cool. Because then you can say, all right, cool. I come in on Monday, and I already know exactly all my funding transfers I need to make for the next two weeks. I trust these. Thanks for the recommendation. Execute. Right. But like, we're not spending a lot of that time on the shorter term, really. Realistically, I know we're looking at holistic forecast and just optimizing ML, but I do question like, are we not. Being aggressive enough on leveraging other data. And I know it depends on customers giving us that data, and on we're doing a bigquery stuff, and, like, I get all that.
Me: Yeah.
Them: But I think that's where. You know, in reality, treasurers, we are looking long term, but our. Our main focus is the next couple of weeks. Of course, right? That's operation. It's like, hey, we need to get this right, because fuck efficiency. If I can't even fund my accounts tomorrow and pay payroll, I don't care how. Amazing. My investment portfolio is for 13 weeks.
Me: Y. Eah.
Them: And that's where they have the magnifying glass. That's where they're really checking. Hey, is this right? Is what I'm being told accurate? We're seeing some of the stuff now. Even with, like, with personio and discogs now, the statement stuff is trying to fall apart a little. Bit or it's not. It's. Something's off. Where? Like the numbers. Like. Hey, wait. I loaded this. I got this. What's going on? That's eroding the trust for them to make longer term, right?
Me: And that's. But that's also the case that I'm making like to have that.
Them: So.
Me: True, like, accurate layer that's not, like, speculative or probabilistic, but a deterministic layer. There's some data that just needs to be very deterministic and always correct.
Them: Yeah.
Me: Is always, like, just a guess because that's.
Them: And that's where. Yeah, exactly. And that's where the connectivity question pops up to me. Because these failures that we're having with a Discogs or the personio is because there's like four links in the chain from bank to palm.
Me: Right. I actually agree.
Them: Again, then the question is okay. The question isn't should we build connectivity because it allows us to execute should we build connectivity because it allows us to build more trust.
Me: Yeah. That I 100% agree on.
Them: You know, so, but.
Me: And also, could it be. Could it. Could it be the case that it's faster now with AI an easier. That it's faster?
Them: It's what? Like building connectivity. Mean, yeah, yeah. I mean, at the same time, we have Fedas, for example. We haven't really, really spent time on. Hey, can. Is this something that's worth pitching or kind of doubling down on? It's more of a safety net right now than it is a default of, hey, look, if your connectivity is crappy, we'll do it for you or whatever.
Me: The drawback. I mean, it could still work commercially as a strategy, right, to sort of land and expand. We can still sell the whole. Yeah, we'll jack into your exist because it's an onboarding like if we're then going to build, you know, so I think I get, you know, you're. Still selling this. I get curious actually selling us doing connectivity already, but it's still such a strong like get through the or kind of. You don't have to go through a three month or six month or a year implementation project.
Them: Yeah, yeah. Y. Eah. It's. It's one of those things where, like. Yeah, it changes. You go to market motion. Right. So sitting on top of connecting with your technology turns into. Well, we could actually service people that don't have EMSes. And now you're going after. I know that it makes it really messy. But for me. What I'm thinking about is, like, the customers that we have, some of the ones that we're talking to, are all like, yeah, we don't really have that great connectivity, you know what I mean?
Me: Yeah.
Them: Maybe it's about, like, okay. Let's be even more. Do we have to be even more? Kind of controlled in terms of like, who is. Who are we really trying to target? You have to have a TMS that has good connectivity. But we're also seeing again with a digital realties of the world. They have a TMS with good connectivity, but it's still not working. And now we're building some sort of connectivity in parser. Right. So it's, it's kind of changing my thought and narrative a little bit around. Like, how much do we blindly trust that if you have a tms that your connectivity is perfectly fine for us, and we're going to be 100% confident with what's coming across because as we're seeing, and maybe it's a, it's the way that we're doing it. Potentially right. Maybe we're connecting to the wrong module of cariba, and it should be a different layer. And. And for, I guess, for digital realty's sake, we're not even doing a connection. We're doing. A. We're doing a data upload, which their data is just that they haven't fixed. But anyways, yeah, now a bit rambling, but that's where connectivity still kind of is in the Is there? In a way. But it's hard to make the commitment because, of course, there's so much better outcomes we could provide than say, hey, we can get you better connectivity now, and then we'll do, like all the cool stuff.
Me: Yeah.
Them: But then it's a circle. Then it's like, well, then how do you build trust if the data is not right?
Me: No. I fully agree on that.
Them: Yeah. Yeah. But in any case, any case.
Me: In any case.
Them: I think it's our job to figure that out. Right? And, and break through some of the noise and have a. Have a conviction and say this is going to be the. Be the path.
Me: Yes.
Them: And then take those steps.
Me: I'm just going to share my latest. I've actually gone through this with Claude a lot, but it. There's an issue with the notion mcp, so it's looking a bit bad.
Them: Awesome. I dropped some notes in here too, yesterday.
Me: I know you did it in a different. This is actually another it put it in. A. I've shared it with you. You should be at it. I'm going to. Yes, I'm going to put it. It's in no slack dm. I've included all your comments so like, they're in the text. It's just not the common signs visible. Thank you for those.
Them: Okay? Okay? Okay. Got it. Yep. Yep.
Me: We like, we can ignore. We can say this. Work in progress, whatever. I just wanted some sort of like, what the hell are we trying to do kind of intro.
Them: Yep.
Me: More than happy to challenge. Turn it upside down, let you know it's more. I think it's important that we can kind of agree on a framing in terms of, like, what we're doing.
Them: I thought this was good. I, I, I don't think I have any problems with this. I'll reread it again just to double check, but, like, overall, like, I was. It was in line with kind of how seeing it as well. Except maybe a few things here and there, but, like, overall.
Me: Yeah. And, like, what do we mean with execution? I mean, like, literally the button, like, not everything leading up to it. Or commodity like it might be. Also that it's so easy to add on at some point that it's not like a factor. But right now, of course, it wouldn't be.
Them: Yeah, yeah, yeah.
Me: Another just. Sorry, Gurjit, but just another quite likely scenario is right if these UIs where you just have a canvas and a chat keep evolving. Like, what is there to stop that from being the main surface of everyone, right? And then you just pull in agents or MCP service or whatever they are called in the future.
Them: Yeah.
Me: And in that context, what are we.
Them: Yeah.
Me: Just as a nice thought exper. Thought experiment. Right, Because.
Them: Yeah. Yeah. Yeah. I think in the end, it comes back to some of the stuff around institutional knowledge, of course. But. It's that. And I still think there is not skiing him back to execution. But then that becomes the thing in a way, right? Maybe.
Me: This idea, like, of just like maybe at some point we can start with. Because it can also be execution, as in autonomous execution, right?
Them: There was. There's.
Me: Because that ultimate, like, trust test, isn't it?
Them: Yeah.
Me: Sorry.
Them: Yeah, yeah. I'll have to show you an article with another treasury system that's being built. They're doing pretty good here in the US Called Treasury Cube, where the founder is saying basically, Interfaces and everything are all going away and that doesn't that like people that are building the pretty interface. It's not good enough now, but whatever, because their interface is also not that great.
Me: I agree.
Them: But what he's saying is. What he's saying the moat for these new systems is the connectivity, is the execution. Because when it comes to a corporation, it comes to what you mentioned on auditability, control and all that is someone still needs to review before release. Someone still needs to own that. Hey, I agree to this. This movement. Someone else needs to control the security of which data is being sent outside of your company to a different platform or to a different whatever. Like that's, that's where the strength comes from or that's really the moats continue to build is if you're that that level deeper of like you're the pipelines, basically more so than the maybe not even pipelines, but.
Me: Yes. I get what you mean. Like the plumbing layer? I think that's the case. Like that. That is still milt. It is still mo. To be that layer. But I think we are also able to add on top of that layer with our amazing forecasting data, like we're creating even more valuable data on top of that basic data. But yes, we could also be a guarantee of.
Them: Yeah.
Me: Like, yeah, they're paying us to be responsible for things being correct. Because highly valuable and nice not having to keep like an individual at your organization fully responsible for. So I think that that that is also an angle I believe personally, at least, to have, like, the accountability. Outsourced in a sense, but anyways, we can. Yeah, we can keep debating this, but I just wanted to see if you agree, like. So what I've added here.
Them: Yeah.
Me: As the potential bet. There, I don't know, somewhere in between features and problems like. Because advanced scenario modeling, like, we can solve this in a multitude of ways, but there's a need for it at least.
Them: Yeah, yeah. Yeah. Like, this is to me, like. So the question I had for you with some of these is that we've kind of already made the. The bet, right? Or we were already kind of saying this is. This is it. We are having advanced scenario modeling. Maybe the question is how much time are we putting into it, but. Some of these. Yeah. Sorry.
Me: Yeah. Yeah, 100%. Like, right now, we're, we're going to release naive version of it and then, but, you know, it's all.
Them: Yeah. Yeah. So to me, like advanced scenario modeling and FX risk intelligence, I don't think that's a trade off, per se, where we have to say it's one of the other. Right. Because it's like, well, we're definitely doing this scenario modeling. We've seen that. We heard that. I know that it's. Going to be core to what we do. So, like, I'm wondering if that needs to. These need to be included as, like, this could put down to that bottom section where it's like the table takes almost in a way.
Me: I can move scenarios to table stakes. It makes sense. I just wanted it to be, like, highlighted that there's a lot more depth here if we want, like. And it actually layers, like, it intertwines with everything else, in a sense.
Them: Yeah.
Me: So. Yeah. Anyways, let's. Let's take effort. Okay, maybe. Maybe. But I think. Yeah. This is actually, I mean, something you brought up, right? So I thought it might. Interesting, because it's also a depth question. We can already now see the FX forecast in your functional currency, and we can build dashboards and we can already do a lot of things. So then what exactly is it that we cannot already do? That would be the bet.
Them: Yeah, yeah. And the funny thing with this one is, like, I don't know if we saw the chat stuff that I put in. As part of the marketing. One of the questions I asked Paul was, this is for some brittle coffee. I said, hey, what are my FX exposures look like if Euro USD go up by 3% in the next quarter? It actually came back with like a pretty solid answer. And then I asked it, well, what if Can I net down some of these exposures across in a company? We came back again with a very solid answer. And, like, it's a place where, like, hey, wait, we could do more intelligence already as is in Pulse. But the question then becomes like, okay, do you want to keep it in the chat, or does it become its own? Is there something we can do on a UI side where it's surfacing that without the question being asked, basically.
Me: So you sort of. You can revisit it without going through the full process. Of arriving there again.
Them: Right. So, I mean, maybe this, again, this is where things just start to overlap. It's like, well, maybe it's a scenario thing where you just type in or you put in, in the scenario, what's the 3% increase or decrease in the Euro USD rate mean? And again, you could, you can see it. I think what I'm getting to is, like, there's obviously different ways to get to the interaction aspect of it now is already starting.
Me: Y.
Them: To be a lot more easier with the polls. But the question then is like, of those outcomes that we know that we can kind of present through, through polls. And in a lot of ways, I'm wondering, like, how do you. Because you. The, the, the UI is not disappeared yet, right? Like, people, there's not many. Companies that are saying, hey, I only work in a chat bot.
Me: O and I think, like you said, there's still going to be a need for, like, quick. You know, quick uis that you, you know, you use a lot, and maybe rather under that chart, you'll talk a little bit more or like, like, yeah.
Them: Yeah, yeah, yeah.
Me: I think it'll be a mix, at least near term.
Them: Y. And so, like. Yeah, yeah, so I'm an FX and. Sorry, go ahead.
Me: Not. I just wanted to say, like, I agree with you, and I see these things a lot, right? He is financer at a small company. Fine. But he did all of this in Claude Coburg. Indirect cash flow statement, okay? Simplified balance sheet, tracking the items that actually move. So ARAP like, did all of these things. Cash, bridge, reconciling P and L to cash. Blah, blah, blah, three to nine. And this is what I was trying to argue in my text that I sent you a little bit like what? What is it that we are building that is not just this general intelligence on top of data? Right. Because this is where I think everything is heading. I think everyone's just gonna use. I don't think for now. I think Palm Chat is brilliant because it's distribution, right? Every one of our users can use it, regardless of their internal IT security. And whatever tools they're using. But these tools, and when I say tool, I mean the canvases and the tools under the surface, like the Excel reading tools. Like, they have a lot of tools built, like the chart rendering tools, like all of these tools that come with Claude or Gemini or ChatGPT, right? I don't understand why why I would switch log into different systems in the future if I can stay in one place. Like, I'm already, like, I'm not leaving my black box. I'm in the terminal a lot because I'm just using cloth code from the terminal, and then it.
Them: Yeah.
Me: 's like.
Them: Yeah.
Me: So what is it that is us? Like what is it? Is it the tooling we provide? We our agents? It's our agent, the internal expert on Treasury. Is it like the decision authority in terms of anything Treasury? So the CFO could sit in his little Claude cowork in two years? And pull in the treasury data and also get like the full context on. Yeah, Paul, the head of treasury, whatever agrees with it or not. So it's like what is it that we are providing that isn't just like entropics? Model on top of data because we're not building the model.
Them: Yeah.
Me: We're building something else. Right. So I think struggle a lot.
Them: Yeah. Yeah y. Eah.
Me: Or I feel like things we can do, but I'm struggling to see that we are aligning in terms of, like, what we are doing.
Them: Yeah. Yeah. No, I think it's fair. Right, so this is like, kind of saying, well, eventually everyone's going to be in Quad co work. And so, like, why would they be in Palm? Or what's Palm going to give Claude co work that no one else can give?
Me: I think, I think there is stuff, right. Forecast, for example, is quite a complex architecture there. And I think like this idea of enhancing or enriching raw data in a sense that is really trustworthy and best in class, I think. And like you say, the institutional knowledge, being able to ask our agent, well, what does our head or treasury think of this? Or, like, what did the team, like, have those, like, decision logs and reasoning logs in the product? So it's not. Just AI thinks this.
Them: Yep.
Me: I think there's stuff we can do like that, but that's so abstract, that it's so new. That it's hard to, I guess, articulate properly. And it's still a bet. It's still a gap. Like, to me, those are the bets. Not like, are we going to do stablecoin? Like that. That. Okay. That. Okay. Okay. Stablecoin is more about. Okay, I give you that. But then I. We could argue that everything on this notion is somewhat.
Them: Yeah.
Me: You know, because bridging. Yeah, they can already do that in Claude. If we just give an mcp, like, Gurney's house for a long time, and he's right, you know, like. Yeah, they can just. They can put that line on top of their forecast in Cowork if we give them the Palm. MCP proper sold, you know.
Them: Yeah. Yeah. Yeah. Yeah. And if they have the cloud, cowork you. Right.
Me: At a surface level, but.
Them: Yeah, yeah, yeah, yeah, yeah. I think the pieces too is like wind again. When does the shift happen where you have all enterprises are in some sort of canvas? I think that's going to take still a longer time than we. Than we expect, right?
Me: Yeah.
Them: And that's where we still have that time to be able to say, well, we're still building as if they're not. And if they. If there are a few that do it, we have that capability, fine, but. I think that's going to take time. But then the question goes, then, is it. Is it. Is it the. Is it the plumbing? Then, you know, is it the. Is it the connectivity? Is it the ability where you're in a canvas and you want to kick out a payment that there's something that's? Going to make that payment happen for you or make that FX trade for you or make that investment for you, because. You're definitely not doing that in Cloud Cowork right now.
Me: No, you're right. Is there an agent for that?
Them: Yeah. Is there an age? Yeah. Yeah. Yeah. But I do get. And I do get. Then why you're questioning the FX intelligence piece. Because hypothetically, as I just kind of said already, you can kind of do this with an mcp with.
Me: Yeah, it's not just that, like, I'm. I'm being intentionally being super annoying right now because I think.
Them: No, this is great. This is great. Like, we need to. Yeah.
Me: In a sense, not really. Here we still. I think we can go deep because if we want to ask, like, What are five top customers? Right. We need to make sure we have that in raw data. So maybe there's something we can do to, like. Get people to that point faster or. Because there's also something there, like, in the plumbing, like, oh, it's so hard to just get the data right. Or get, like, transactions that are properly categorized, for example. I think that's a great example. AI has made getting properly categorized transactions. Easier. But yeah, one could just argue that.
Them: Yeah. Yeah, again, for something like the five PIN customers and the raw data. Then again, where does that raw data sit? It's typically going to be in your bank statement or your ERP system, right? And. The DSO DPO, that's going to be in your ERPs as much. So then it's kind of like. Again, are we. Then you have your interaction layer, which is whatever canvas you're using. Then you have palm that has your. Your institutional knowledge that you've shared over time. Has the context, which is, I guess, pretty much the same. And then it's also what cleaned up data that's actually being pulled from a tms, an erp, whatever, payroll system, CRM, maybe even. Is that kind of where we sit, then? Is that we're that little. The thick middle you're talking about?
Me: Or are we, like you say, building our own connectivity and we get the raw statements and we can see the descriptions and make our own best guess on this and then be very transparent. Hey, this is our best guess. And go speak to your art team about better descriptions.
Them: Yeah, yeah. Yeah. Yeah, I kind of. I was trying to draw it out. I'm looking at my finger on current, like, what is a system of intelligence? And the way I kind of. Drafted this piece was like. You know, you have your ERP that's going to show you your expenses and maybe some revenue. Your CRM is also showing you revenue or whatever your sales teams are putting in to, like, expected deals to close that's going to eventually, at some point, hit into your actual cash. You have headcount in your. Your HCM or your headcount management tool like workday or whatever. Then I put TMS down as you're. As your cache data database. That's where your cache information is. Your bank statements are. I mean, that can be switched out for tms, or it could be a direct bank connection that we build, whatever, but it's still, like, it's still one of four. Other pieces, and I'm sure there's a thousand other pieces that are technically also influencing cash. In a way, right? And then the. The ability that we have. I think the opportunity that we have in front of us is how do you get those different. Wherever those different silos of data is, and all of those little silos are, in one way or another, influencing what's going to happen with literal physical cash at your company. Is, can we become that tool that's able to now query and understand because of all of these nuances that are happening and all these other databases around your company or your systems of record throughout the company? Is actually taking that mess and then figuring out, okay, based off of what's happening here, this is what is going to happen in the future. Whatever. Like, that's where the forecasting comes in. But also, it could be just overall intelligence around, why is Cash behaving a specific way? What's changed in the past. What's. What's changed today that wasn't expected. Wasn't expected yesterday? Like what's driving all of this change, right? Just an example would be like, okay, say somebody closed a huge deal in the CRM and the contract says that it's going to be. Payment is due on signature. I don't know, making something up, right? When typically everything else is on contract based. Well, hey, does that now that actually changes the way that we expect money to move because it's not based on, hey, trends or whatever, is that this one deal is a large one and we're going to get paid tomorrow and blah, blah, blah. I don't know. Just as an example, but you have all of these things in these different sources of record across the company that are happening. And happening every day, and they're all impacting cash at some point. Can we be that kind of top piece as it will just go in and understand based off the context we already have from you as a Treasury and the data that you have within the company and smash all that shit together and just be able to say, look, here's a confident way around how do we think about your money, whether it's how much you have, how much you need, how much you can invest, whatever. And I think, like, having this canvas on top that does that. I don't know if cloud coworkers connecting into ERPs and CRMs, and maybe it is at some point. I mean, it, it could be, but.
Me: No, I don't think there's going to need to be. At least for the very first, like, foreseeable future. Some, I think, one could argue, like, everything will just converge. Like, why will there be different systems if it's so, like, that's one argument, right? Why would you need a CRM? And blah, blah, blah. Like, why would you even need that? Maybe. Maybe. A big technology giant drops a product that is just like everything. Organizational context, agents. Now, I'm way ahead, but, you know, like, automatically setting up integrations for you, ingesting your staff so you don't need differences anymore. But I still believe, like, There is a point in going really, really, really, really deep and being the internal expert, like, in a specific domain.
Them: Yeah, yeah. Yeah. Great. So that's what. That's what I think. That. That's where. That's where opportunity lies, right?
Me: But that's why I like thinking about it as the agent. Like what if Palm is an agent, what role would it have? Like thinking about Palm. Almost like a human. Like for human. Now I'm going. I don't know if it is. If. I don't know if this is vibing with you. But.
Them: Y.
Me: If Paul wants superhuman, what would Palm's capabilities be like? What would Palm freaking do at a company?
Them: Eah. Yeah. So let's. Let me. Let me flip it a little bit. Like what? What? You bring in a new analyst and you got a headcount now, and you're like, okay, we're bringing someone in. What jobs are they going to do? You know, and depending on how smart they are or whatever. But we want them to do everything. What does this person do right? Like, we, we've done a lot on the cash management side. This is forecasting and liquidity management and making the decisions on when to fund, what to fund, blah, blah, blah, investments, fx, all the stuff that we're talking about here, right? But a treasurer is a Treasury analyst or a Treasury person's day is not literally just doing cash management eight hours a day, 40 hours a week. Right. So you do that, and it's typically your first kind of first half of the day, maybe, maybe, maybe even less, depending on how you're set up. Then you have quite a bit of other things that happen that are kind of outside of cash, but they're treasury focused, which is things like bank rationalization, opening and closing accounts that are not necessarily needed. Or you're doing, you're doing great confirmations, which are part of an annual audit, and you talk literally just emailing banks back and forth all day just trying to get information or give information. You're doing kyc. You're doing a weekly intercompany. Sorry, a quarterly intercompany settlement that takes a whole day. That takes one whole day to do. Or you're rolling your hedges, which. Is one specific day. It's not every day, but it's one day of the month that you have to do it. You have your reporting we already know about. Then you have. You have things like bank fee analysis that you're doing, and you're trying to run a project to run. Okay, how do we get our bank fees lower across all of our banks? And where do I pull that information to know, am I overpaying? Am I underpinned? Can I benchmark? It somewhere. And so you're doing these other kind of things outside of just strictly cash management. That all rolled into cash management, obviously. Right. So that's kind of the core function. But if I had an agent, some of the things that I hear a lot about or people are saying, I don't think we go this way right now or maybe ever, I don't know, but, like. That Uber 18 people. I don't know if you're the this if I shared about the at the dinner where so at Uber, we had outsourced to Accenture to BPO some of our treasury op stuff that's like, low value and, like, but it does require someone to do it. Like, yeah. So downloading bank statements.
Me: Who's the people in India? Manually.
Them: Yeah. Confirming payments have settled some of the reporting that's around that, right? Like, they went from three people when I left, 18 people now. And, like, the question is, like, well, can an agent do a lot of that already and faster?
Me: Yeah, I heard.
Them: And now you have 18 less people that you've outsourced to to do something that an agent can do. That's not cash management specifically, it's a Treasury function, some cash management function.
Me: But this is. Yeah, but can I just say before I forget, because the other big trend, right, if you look at B2C with AI and everything, is that people believe in the, you know, the whole hyper personalization angle. That, you know, a year or two will all have our smartphones with our own model, and everything's going to be super tailored towards you.
Them: Yeah.
Me: So I think the same will be true. It's going to be something that counteracts a long held belief in the sauce space that you cannot build. Custom problem. You know, I think this is very interesting with agents and, and how can you tailor them then in these specific ways for your setup for Uber, like, okay. So they have that thing.
Them: Man.
Me: Too bad. I'm feeling bad about the Indian guys, but, like, honestly, that's probably.
Them: Yeah, yeah, that. That was a question I had, actually. You know, I'm glad you brought it up. It's like, with the cost of development going down,
Me: You know? Y. Eah.
Them: Can we really push for, like, hyper customized treasury tools per customer?
Me: I don't know. I hope. I would love to think so.
Them: You know, and so then it gets to a place. I mean, we're doing some of it for like year old ports now. We're trying to build a few things for them specifically. Then that. That becomes a whole different.
Me: Exactly. Because we can argue already we're doing a little bit customization, right? A little bit in the forecasting, a little bit around transactions categorization. We are sort of taking your organizational context into consideration.
Them: Yeah, yeah. Yeah.
Me: But we are not going hyper customized like super. I think this conversation will scare Rodal, but I think I think that's an interesting.
Them: Yeah. Yeah.
Me: Track to explore.
Them: Yeah, yeah, yeah. Because again, like, it's just. It's kind of a mental shift, right? We're all used to SaaS. Model of, you get the tool as it is, you customize you. You form your workflows to fit into what the SaaS offers. Because everybody gets exactly the same thing. Maybe. Different modules, but you get the same thing.
Me: Because it's been. Freaking expensive. Yeah. To build.
Them: Exactly. Yeah, exactly. But if we're going to a place where one is ui, UX is just changing in general. Some people might want interfaces, some people might not want interfaces, some might want outcomes, some might want to focus on just one thing. But on top of that, layering it onto the institutional knowledge, the way that your company operates, what's important to you, that's all you see and nothing else that's irrelevant or unimportant. And you can ask for more of the important stuff you need. It's easy to kind of bring into the loop. That's obviously super interesting. I think there's questions around maintenance and all that stuff, too, though. And where does that go and what does that look like?
Me: I mean, I think if we even started at the whole institution, a knowledge layer, like, capturing these, like, quirks that are typical for your organization and your policies and your, like, things that you care strongly about and how. Yeah, I don't know. I think we still need to even explore what that is.
Them: Yeah.
Me: That's one layer of hyper personalization, right? But we need to deliver on it. We need to make use of it. Then we need to make sure that if I'm in Palm Chat, It's proactive in saying, hey, you know, but think about your policy now, and this might breach it. Look at like, so how do we design those experiences? And this is why I'm so excited to get a designer on board that's, like, really into systems workflows like that, because at the end of the day, That's what matters more than I agree the pretty UI or like those sort of things that will but But I also wonder. Yeah, we can provide the ui. But does it matter? Or will that also just be so easy at some point for someone like, will there come a tool that is just like, yeah, your, your dashboards connect it with any data and then you can just have your UI layer here on top of your 85 agents that you use.
Them: Yeah.
Me: But for now. I know, I know. I'm. I'm. I'm stretching it, but for now, we definitely still need the app, like, the web app when it's going to be a transition, but, yeah.
Them: Yeah. Yeah, yeah. I think you wisely still matters. I think it will continue to matter. I think there's still kind of ways corporates work and the way that they're sending their information across to different departments, there's a reporting aspect of all of that. At the end of the day, like, there's still, like, a decision.
Me: Y.
Them: All of this needs to lead to a decision. And then something happening with that decision.
Me: Es. But something I got looked up on. Sorry, Gurjit, but, like, well, always been this, like, element of so reporting. It's so broad, right? It can either be hated, quick lumber, quick insight, quick. What is that like? Or it's essentially something I'm sending away to my cfo. Or it can even be. The CFO himself going in because now he has access. Tom mentioned it. In a clip that Maria shared. Like, amazing. Now my SIFO can just ask for the data. He wants himself. Like I don't need to give it to him.
Them: Yeah.
Me: So watching, and it's becoming more collaborative. But also the. The thing I found interesting that's always been a thing is, like, the whole storytelling.
Them: Yeah.
Me: Aspect. Right. And what can an AI like, what could we do to help tell a story instead of a PDF?
Them: Yeah. Yeah, yeah.
Me: That may be the reporting we should build instead of copying embeddable with. You know, it's just like this is why I go back to users and say yeah, well what else? Like okay, sure adoption is not great and self serve, but I'm sure the use case I just want to quickly check something they can check it in the app. We can show a chart in the app. Great.
Them: Y. Eah. Yeah. Yeah.
Me: But in this case, I really want to get by in internally or I want to tell a story about this, or you know, that we can do that better.
Them: Yeah, look at, look at the euro ports. You probably already saw it, right? The PDF that they shared at their. Their board reporting. Yanis shared it, but that's, that's the story. Sorry.
Me: No, I. I missed the board reporting PDF.
Them: Yeah. So that's basically what they're trying to replicate in Palm. And it. What it is, is you have your chart that shows your last year's actuals, this year's actuals, the forecast and budget, like four lines. Fine, that's the chart. But beneath it is an actual narrative on why is it different? What's. What's caused it to be different, what needs to change for it to be better, whatever. Like, so that's kind of what, when I think of reporting, that's what I'm thinking is can you build these leadership decks, which. Is not just a chart for someone to figure out, but it's a chart that visualizes a story which is told at the bottom, which is also generated. Could be user, it could be AI. Could be what? A mix of both? Because, hey, maybe there's something that you want to also include. But whatever. But that gets to a point where now you're saying, hey, my board deck is created in 30 seconds versus a week and a half of work and has the narrative it has a story aligns with what's important to us as a business.
Me: Exactly.
Them: Maybe they don't care about fx, Maybe they care about working capital, whatever, right? And so Palm knows that, hey, their narrative is based on working capital management. That's the main focus on these decks. So here's the chart and here's a story on how's working capital at ports or whatever, right?
Me: So stupid idea. Maybe. But wouldn't it be cool then, too? Because we need to find out where we're lacking. Let's say the working capital. I want to create a story, a strong story, and, like, just report on where we're at there. What's our dies. El, you know. Wouldn't that be a fun internal experiment and try that out on each of our customers data. And see where it breaks, where it just, you know, starts making shit up because we don't have the proper raw data.
Them: Yeah. You mean like checking in pulse, for example. Like, just type it in there and just saying, hey. Yeah. Yeah.
Me: We could do it from within, Claude.
Them: Yeah.
Me: Right, so just get the charts, get everything. Because I think that's a good way for us to stress test, like, what capabilities do we already sort of have and where do they break? And then it's easier to scope out, like, understand the width of working capital, intelligence. I mean. I have no idea the capability. I think. I think it's probably would be nicer if let's say working capital user asks about that. We have a tool behind the scenes that really, really good at where agents that are really good at working capital, they know exactly what data to pull. So it's always the right data. They know exactly, you know. So I think there's stuff we can build around it, but it would still be interesting because.
Them: Yeah.
Me: Yeah, I'm also a little bit like. Yeah. The scenario prototype. It's slow. The UI here and there. It's probably pretty fast to just do with AI on a canvas.
Them: Yeah. Yeah.
Me: Tooling that saves the data and. Yeah, whatever.
Them: Yeah. But I think again, I think, at least for today, I would say hardly anybody. Is using any sort of. I mean, you hear about copilot, maybe here and there, but there's no one that has their work, no treasure that I know that has their workflow.
Me: Yeah, I know, I know. But Johnny said unused Gemini internally, but it's still unclear how much.
Them: Yeah. Yeah. I mean, because then on. Then on Gemini Kariba palm.
Me: Yeah.
Them: Yeah. I think I even sometimes still struggle around the lack of. If we get to a place where there's just nothing to log into, it's just like a. You have a empty canvas and you just start asking questions or it pings you some things that are important. Like.
Me: But that's. I think, like, maybe. That the sign comes in, like, right? The just goes, hey, Palm. And then we are proactive. Yeah. Do this. Talk about this like, oh, yeah. Like, if we're always expecting, like right now, this user experiences are always like, oh, user has a precise question that they want to know. And, you know, maybe it can be more proactive also in those flows.
Them: Yeah.
Me: Again. This is a design domain thing that needs to be cracked, but. I don't know.
Them: Yeah.
Me: I'm happy to discuss this Wednesday, but there is no. I just wanted to say like. I have just added a little bit of like for each one.
Them: I saw that. Yeah, yeah, yeah, yeah, I have.
Me: Simple enough evaluation. Like I have no idea. We can just guess stuff. Pat to Revenue. I also think is quite interesting because we already have some like.
Them: Y.
Me: Can we add on to existing customers? Is it something that will definitely land us a different type of, like, I don't know.
Them: Eah, yeah, yeah, yeah.
Me: And then this is the part.
Them: That's on me. Yeah? Yeah. So. I yeah, I have a yeah, I have I have plans to go through this and then fill in those pieces. Like, I'll I'll definitely pick at a competition stuff. I could talk to Path Revenue at least what I know from kind of at least the industry, at least or some of the customers that we've spoken to.
Me: Sorry. Just quickly. Gujarat. I'll remove this one or add it on the table stakes.
Them: Yeah, I think so. I think so. Fx, Risk intelligence. Now, I'm questioning some of this stuff, too, after this discussion around, like, is this. Is this really a bet? Bet, Right.
Me: I mean, hey. Where did. Oh, yeah. It's broken. I used to have you. You added a long list of stuff that I had here before, but whatever. In another place, but yeah.
Them: I think let's just leave that FS risk intelligence working capital intelligence, I think makes sense. Direct and direct decision.
Me: But I think I'm wondering if this is like a huge. Like, this. To me, this just feels more and more like. Like what you. Or it's. It's essentially like. I think it's something like this, so. They want to see variances and then maybe drill down a little bit into, like, the palm categories.
Them: Yeah, yeah.
Me: And then that helps explain a lot. It doesn't help a lot in terms of working capital, but it. It can surface some other things.
Them: Yeah, but this is the, this is where, again, it's all this stuff overlaps so much, right? Because, like, you have, you have this, which overlaps with your kind of like, working capital stuff and DSO dpo potentially. Because what you do is you adjust your DSO DPO to get this back on track. If it starts going in the wrong direction, like, so you pull that lever to correct this lever, and then correcting this lever will then impact your fxp, you know, so it's. That's the part where.
Me: It's so hard to do, but I still. We could look at it purely from a data perspective. Lens, maybe.
Them: Of where we have the access to. Right? Like where what can we already surface?
Me: What can we already see? Can we even, like, do? I don't know. I was playing around. I don't have time. I never have time. But I would have, like a tool just for us internally to try out in Claude or whatever. In our mcp. I mean, sorry.
Them: Yeah.
Me: That supports more general file uploads into Palm.
Them: Yeah, yeah, yeah.
Me: Because if we can make that happen, We can make a lot of these things happen.
Them: Yeah. And then you could just start seeing what happens when you drop in a budget. And how does it. How does it Think about it, right?
Me: Anywho.
Them: Okay. Yeah. Decision and governance, intelligence. I think this to me also is like, I see this ender table stakes, to be honest. And like, I guess the question here is also similar to scenario planning is how deep do you want to go? Right. And that's maybe the more strategic question is how much time do we keep? Spending on on this.
Me: I rephrased it a little bit because it's like, to me. It's maybe even just this, to be honest, but because it's very. It's for just like payments or transferring, like cross countries. It's like there are a lot of use case, use cases.
Them: Yeah. Yep, yep.
Me: I don't know why it's compacted everything so much. I had really good. Convict here, but it's. I don't know.
Them: All good.
Me: Just the deeper IC intelligence.
Them: Yes. Yeah. Yeah, I think I would call this more into, like, just. I guess we could leave that deep, deep, deep, Right. Nice intelligence. But this goes back to kind of the differences of intercompany, right? Like there's stuff that we're doing, it falls under something else. Like pooling and stuff is something else. But then if you talk about in house, bank collections and payment, on behalf. That's something definitely different in those are different. But I guess that's the question here, is like, do we go deeper on those other things, right?
Me: Yeah. And also, like, in house banking is one thing that I just put out there because it's like. Like, if we talk to someone like Levi's, at least that's.
Them: It's huge for them. Yeah, yeah.
Me: But I'm also not an expert. And in house banging,
Them: Yeah, yeah.
Me: To me that borders, or at least tightly connects to intercompany.
Them: Yeah, yeah. And then you had connectivity as well. What I'll do is feel free to update whatever you needed to do still. And then this evening, I'll put time into it to fill in the stuff around the revenue and competition, and then if I have any other thoughts, I'll. I'll drop them in there. And then. So then by tomorrow we should be able to have.
Me: What? It kind of makes sense to keep. We keep direct. So I'm just going to. I think I am having Granola. We keep direct bank connectivity. Because I think that's interesting. We keep deeper in icy intelligence. We keep this one? The stable, coins and stuff. But. This one might be table stakes. The decision and governance. It's something we just have to do.
Them: To do? I think so.
Me: It's something that we should highlight as we should spend more time here, but it's not really a bet that we should discuss.
Them: Yeah, yeah. Y.
Me: And then I think maybe the whole. Maybe I can just put the direct indirect bridging in under working capital in tallies. I merge these two.
Them: Eah. I think.
Me: Because then we can discuss more about how deep we go. Because we can do like super surface level. Yeah, they can see FP and a overlay on there or whatever. Global forecast. But if we want to go deeper into helping them understand what is driving the differences between the two,
Them: That's working capital.
Me: Then we're going to need to look into, for example, working capital and definitely go into ARAP and all of those domains even deeper as well.
Them: Yes. Yep.
Me: Which what customer behavior is driving this and blah, blah, blah. So I think to me,
Them: Yep, yep.
Me: To me. This is.
Them: I think that's fair.
Me: I think we can just say, hey, yes, level one is just a bridge having something that bridges. And then level two is really, really explaining the differences.
Them: Yeah. And then getting to a decision like, what do you do with these differences? And what are your options?
Me: Yeah. We'll keep ethics, risk and talent. I think still we can still discuss it. And what it means or to go, like, really deep. Because if we are going to position ourselves as an FX tool, I mean, that's got to come with something more. Than what we currently do.
Them: Yeah, yeah, yeah. But I. I would say, like, even on just the intelligence piece, There's, you know, companies that have been built around just that is showing your exposures. Right. And I know technically you might seem like it's easy, hey, we're just taking data and presenting it to you. But I think this is where I would keep in mind that for the treasurer themselves is the way that they're doing it today. It's, it's. It is painful. And if it's easy to build and it's. It's useful for them. Trust aside, like trust is again. Say the trust is there now. Hey, if it is easy to just literally be able to present to you your exposures of the next couple of weeks or months, that's already pretty powerful.
Me: Yeah.
Them: Even if it's technically easy in, we're like, hey, that's not that hard to build. That's great then, for us, right? Solving a hard problem with the easy solution. Never a bad thing, I guess. But then there is, I think, on the intelligence side, though, at the end of the day, Where the depth comes from is going into the decisioning aspect of it. It's like, okay, here's what you have. But here are your options and decisions you should be making that are going to impact your business, right? And then you have actual hedging and hedge accounting, which is, like, super complex. And it requires a lot more data from ERPs and from accounting teams. And there's, there's a whole, that's a whole beast in its own. And then how do you kind of measure the effectiveness of your hedges that's where I think the complications will eventually come. But it starts from versus knowing your exposures. And so if we're saying we're building an exposure management or intelligence, next up is then exposure decision and then execution and then measurement.
Me: It becomes a bit similar to this piece. Like, yeah, we might do the bridging, but then, like, what's the next. Deeper, I think. And we're just going to have to live with. Some of these bets are more like, how deep do we go when? And others are, do we even do it like this? One, for example.
Them: Exactly. Yeah. Yeah. Yeah. Yeah.
Me: Seems like a bit of a buzz around this in the treasury world, so why not?
Them: I'm going to. I'm going to a seminar next Thursday that has people from Stripe and Circle. A couple of treasures are going to be there to talk about the impact of stablecoins on Treasury. I'll take copious notes and share what I hear back, but I'm really excited for it. Because I want to hear there's gonna be treasurers there. There'd be people that are actually using this now.
Me: Oh. Wow.
Them: So later than this session, but at least I'll have some more more to share on that one.
Me: Very cool.
Them: Session. Cool. Okay.
Me: I'll try and get this in better shape as soon as I can.
Them: Okay. Awesome. And then I'll spend time on this evening, and then we'll be ready for tomorrow.
Me: Super cool. It's going to be a fun.
Them: Awesome. Thanks. Thanks, Emma. This is going to be great. Appreciate the work. See you. Bye.