Skip to content

ON - Batch Upload & Forecast Settings Prototypes - 2025-06-26

Metadata

  • Date: 2025-06-26
  • Company: ON (On Holding)
  • External Participants: Amanda Mitt (Treasury), Rodrigo Cabrera (Treasury), Yulia Ershova (Treasury)
  • Palm Participants: Christian Sobkowski, Emma Sjöström
  • Type: Customer Call
  • Domain Areas: Cash Forecasting, Categorization
  • Recording: https://tldv.io/app/meetings/685cf7d4e597140013fa48e1/

Summary

Context

Regular customer call with ON treasury team. Emma presented two UI prototypes for feedback: batch upload of forecast data (payroll, tax, etc.) and account-level forecast settings to control ML vs manual input per category. Amanda also raised contract renewal discussion - Christoph wants documented use cases as they approach end of 6-month period.

Key Discussion Points

  • IC Update deploying - intercompany categorization improvements being loaded, including cash pool interest
  • Contract renewal - approaching end of 6-month period; Christoph wants use case documentation; Amanda creating problem/solution/next steps table
  • Batch upload prototype walkthrough:
  • Upload spreadsheets (payroll, tax, etc.) with any column format
  • AI/LLM automatically maps columns to Palm fields (date, amount, account, currency)
  • Column mappings remembered for future uploads of same file format
  • One category per file initially (future: auto-detect category)
  • Row validation: green (valid), yellow (duplicate), red (invalid/missing data)
  • Download invalid rows as CSV in original format to fix and re-upload
  • Visual distinction in forecast view for uploaded vs ML data (gray boxes)
  • Tax payment tracker & talent payment tracker files reviewed:
  • Payment due date vs actual payment date
  • "ASAP" dates problematic - could default to next Thursday (payment run day)
  • Rodrigo showed actual file structure with tabs, columns
  • Account-level forecast settings prototype:
  • Turn ML on/off per category per account
  • Options when ML off: Default to zero, Manual input, Fixed recurring amount
  • Global vs account-level settings (Amanda: "global would make sense")
  • Combination forecasts requested: ML + ERP data for AR/AP

Pain Points

  • ASAP dates - some payments just say "ASAP" with no actual date, need logic to convert to valid forecast date
  • Kyriba/SAP inflexibility - "In Kyriba or SAP, it's always a lot of work to update all the forecasts... you always have to take it back to Excel"
  • Working on categories before variance analysis - can't tackle variance analysis yet while still fixing categorization

Feature Requests & Needs

  • Batch upload with intelligent column mapping (validated - "exactly what we needed")
  • Download invalid rows in original file format for easy fixing
  • Global forecast settings - set payroll to always be manual across all accounts
  • Combination forecasts - ML + ERP data together for AR/AP categories
  • Edit dates directly in UI - fix ASAP dates during upload rather than re-downloading
  • Duplicate detection against ML predictions - not just manual duplicates (future consideration)

Jobs & Desired Outcomes

Job: Upload forecast data from team spreadsheets without manual data transformation

Desired Outcomes: - Minimize the time required to map spreadsheet columns to forecast fields - Reduce errors from manual copy-paste between systems - Increase the number of forecast data sources that can be easily ingested

Job: Control forecast methodology per category to match data quality and availability

Desired Outcomes: - Minimize reliance on ML predictions for categories with known future data - Increase flexibility to use manual input, ERP data, or ML as appropriate per category - Reduce effort to update forecast settings across many accounts (global settings)

Domain Insights

  • Payment runs are weekly (Thursdays) - even "ASAP" payments wait for next Thursday
  • Tax payment tracker structure: Payment type (actual/refund), account, amount, currency, payment date, due date
  • Talent payment tracker: Covers non-salary HR payments (expenses, reimbursements)
  • File layout stability: "I don't think they've changed that often... new tab every year"
  • Forecast source preferences at ON:
  • Cash-ins: ML + maybe assumption for growth
  • Tax: Manual
  • Salaries: Manual
  • Fees (RCF, etc.): Manual
  • Open AR/AP: Combination (ML + ERP data)

Action Items

  • [ ] Christian & Amanda to meet tomorrow for 30-min use case documentation session
  • [ ] Amanda to share ideal list of which categories should use which forecast method
  • [ ] Emma to continue batch upload development (1-2 sprints)
  • [ ] Palm to notify when IC update is ready for testing

Notable Quotes

"I think it's exactly what we were... the need we had. And I like how it would be easy to track what was manually... via spreadsheet." - Amanda Mitt

"In Kyriba, or in SAP, whatever. It's always kind of very... it's a lot of work to update all the forecasts and like just yeah, you always have to take it back to Excel in the end. So I think it's nice that you can see all of it in the tool." - Amanda Mitt

"I feel like we're getting where we dreamed of." - Amanda Mitt

Full Transcript

Date: 26/06/2025, 09:33

00:00 Christian Sobkowski: and getting feedback, unless there's any, any points from your, your site that you want to chat with before,

00:00 Amanda Mitt: No, I know we wanted to discuss the IC. Update. And also, Yeah, I have a point but we can leave it for the end.

00:02 Christian Sobkowski: Let's, let's do. Let's get it out of the way now, so quickly on IC Update team is loading right now. So if something changes in the app, that's that's us reloading. I want to run two, three tests. Before before handing it over but it's on the way Amanda.

00:21 Amanda Mitt: Okay, thanks. Yeah, I already noticed some updates like cash, pull interest.

00:23 Emma Sjöström: Is.

00:27 Amanda Mitt: but yeah, let me know when

00:28 Christian Sobkowski: Okay. Yeah, I'll let you know, I'll let you know.

00:29 Amanda Mitt: Ready.

00:33 Christian Sobkowski: What's what's the other point?

00:35 Amanda Mitt: Now, it's just because we've been as, you know, we Are reaching kind of the end of the six months. We agreed on.

00:44 Christian Sobkowski: Yeah.

00:46 Amanda Mitt: And yeah, of course, we want to continue but

00:50 Christian Sobkowski: It's nice.

00:51 Amanda Mitt: Christoph. Always likes to kind of challenges a little bit and he wants us to

00:54 Christian Sobkowski: Of course.

00:55 Amanda Mitt: really all of the use cases that we need palm

00:58 Christian Sobkowski: Yep.

00:59 Amanda Mitt: Too. I started so maybe I can share but nice to get your view as well and how

01:02 Rodrigo Cabrera: which,

01:03 Emma Sjöström: Yeah.

01:06 Amanda Mitt: you see this? Use cases how maybe we can. Yeah. So all the value that the two is bringing to us. I also think a lot of it will come now because we're integrating liquidity planning in caliber So, for example, I think Palm will help a lot with variance analysis.

00:00 :

01:19 Christian Sobkowski: Yep. Yep.

01:23 Amanda Mitt: That's something that we're not super able to tackle yet because we're working on the categories. So yeah. just wanted to kind of pair this document together with all of you with all of

01:31 Christian Sobkowski: Up.

01:35 Emma Sjöström: these Sounds great.

01:38 Amanda Mitt: I think would be nice. Also, maybe map a little bit more the technology involved

01:39 Christian Sobkowski: Yeah.

01:42 Amanda Mitt: Um, LLM or, I don't know, like the more historical analysis or Regression. But let me share quickly, I was just starting to put the table like

01:58 Christian Sobkowski: Very cool. Yes.

01:59 Amanda Mitt: this. so, just by problem solution, and next steps kind of

02:07 Christian Sobkowski: I,

02:07 Amanda Mitt: The problem, of course, I don't, I don't think it's like a problem that we still have it now. But that was why we started working with you guys in the first place. So for example, we had no Farquest process in place. And clear cache needs per account and the solution was to build an automated forecast process in the next steps that we're working on.

02:25 Amanda Mitt: So maybe that's what you're going to share today and a lot of next steps.

02:29 Emma Sjöström: Yes, don't, exactly. I'm gonna show you the first iteration of how we'll get

02:30 Amanda Mitt: it's just like, Yeah, I think it will help a speech to Kristoff because of

02:33 Emma Sjöström: more data. In.

02:39 Amanda Mitt: course, we want to continue but,

02:44 Emma Sjöström: this is,

02:44 Christian Sobkowski: I appreciated, Amanda.

02:45 Emma Sjöström: This is great.

02:48 Christian Sobkowski: so sorry, let's do how Would should is it helpful Amanda if you and I speak for 30 minutes?

03:01 Rodrigo Cabrera: which,

03:01 Amanda Mitt: About this. Yes, I agree. I think today, we had a lot to impact, so maybe we can have a session.

03:08 Christian Sobkowski: Yeah.

03:09 Amanda Mitt: Tomorrow, I'm pretty free.

03:11 Christian Sobkowski: Amazing, feel free to. I'm free all day tomorrow. Feel free to book me and we

03:11 Amanda Mitt: If you want.

03:14 Christian Sobkowski: can go through together.

03:18 Amanda Mitt: Okay. I'm sorry, the chat.

03:19 Christian Sobkowski: Super nice. Thanks. Thanks for being super active in this.

03:24 Emma Sjöström: Yeah, thank you. All right, curious to learn more about that also. But it looks, it looks good at a high level. I have to say, it feels We're wrong. So what I'm going to show you today are two things One. And okay, remember first of all, these are just kind of high level UI prototypes.

03:45 Emma Sjöström: So they're not functional, it's not something we have in the app yet one of them we've already started building. But I just want to sanity check some things with you because I really want to make sure that you will find it usable. Even in the first version, right. So, the first thing I'm going to show you is kind of a batch upload of Manual forecast or like it could be payroll or taxes.

04:14 Emma Sjöström: Those sort of things that typically live in spreadsheets here and there And the other one, I want to start exploring with you and just getting your early feedback on, is being able to control at the account level, sort of turning on and off machine learning for a specific categories.

04:33 Emma Sjöström: And if you turn it off, what would you like instead? And so that, that would be also really valuable to get your thoughts.

04:42 Rodrigo Cabrera: so,

04:43 Emma Sjöström: How it would be usable for you guys. So what do you need to make it? Something you can use even if it's not perfect at once and then, we can also discuss like how we will iterate on it, and make it better. okay, so just to make One thing that could be good to know.

05:09 Emma Sjöström: Clear. First Is this is the current up, right? So, We have. Support for manual input or kind of one-offs and using this action here. As, you know, And I know you've been using it a little bit, so also curious for any comments or thoughts here if it's helpful or not.

05:33 Emma Sjöström: because in the first, Version of the kind of batch upload feature. We're building, we're going to keep using this UI just initially. So your uploaded data will show in the same way like These. Kind of one of. Shows up just so you're aware and please, any thoughts is great.

05:56 Emma Sjöström: But let me just show you quickly then how we're imagining the first version of this patch upload feature to work, it would allow you to upload a spreadsheet. What I'm using right now I'll show you. It's just some example data. it's actually this file with just some payroll forecast with some, you know, columns that are named Whatever.

06:24 Emma Sjöström: but the idea is we are going to In this first version. Have you set which category applies across the whole file? Okay, so we'll just be in the first version. One category, profile. Okay, so this is payroll. Then I choose payroll. and then the idea is that we will Automatically.

06:48 Emma Sjöström: Kind of map, each column in this file. To the corresponding concept of a forecast in Palm, right? So we will identify the the date, the amount account currency. We will give you the opportunity as well to adjust if the model got it wrong. But then the idea is have you.

07:12 Emma Sjöström: Uploaded a file. Once it will remember those colon mappings. So you wouldn't need to do it again. If the columns are still named the same, our system will recognize it and just immediately. Map it like this for you. You'll still have the option to edit, of course, but That's the idea.

07:36 Emma Sjöström: So you don't have to You know, no matter the file format, you could kind of just have this automatic mapping. To the right column. So then what we imagine is that you would have rows from your file that are perfectly valid and that can go into the focus straight away because they have all the fields required.

08:03 Emma Sjöström: We also imagine that there might be a potential for duplicates. Perhaps you've added something. Other one off or perhaps you've. You're somebody else. Your colleague has intruding ingesting the same file already. So it's already there. What we just wanted to make sure was to draw your attention to if there is already for the account, the amount and the date, something that exactly much what you try to upload will make you aware of it.

08:38 Emma Sjöström: And then you can choose to optionally Include into the forecast, anyway. And then we're also imagining that inevitably every now, and then there will be some rose type. we for one reason or another cannot ingest for example, if there is no date or amount or no way for us to automatically map the Account numbers, right? So, what we're imagining is that we would allow, for example, your Creeper account number.

09:12 Emma Sjöström: So yeah we'll create nothing. We have mappings of your account numbers to palm account numbers so that would be automatic as well. and just so the last thing we were thinking here was also Well, okay, so these rows are invalid. We would let you download a CSV. Using the original format of your file.

09:37 Emma Sjöström: so this file format with all these columns, But only representing like these rows, that were invalid, so that you could have a nice way to to know that. Okay? I need to reach out to so and so to get this number and then you can just re-upload that file if you want.

10:02 Emma Sjöström: Optionally. Of course, you can always just re-upload the original file with updated numbers in terms of the rosewood that we're missing a number and you would end up here but with a lot of duplicates and like just a few green ones for example. But yeah, just selecting the yeah.

10:26 Emma Sjöström: You choose which ones you want to upload and then you click Confirm Upload and then it goes into your forecast. And it will show up like this in the first iteration. So this gray boxes indicating where there's Kind of manual distinct values that has been uploaded. What are your thoughts? And top of my

10:55 Amanda Mitt: Cool. I think it's exactly what we were the need we had right to And I like how it would be easy to track what was manually kind of not manually but via spreadsheet. Yeah, looks exactly what we needed. So Yeah the duplicate I think it's helpful as well in case there's sorry any historical similar payments coming on from the model from the AI.

11:19 Amanda Mitt: Are we able to see it in that case? No, right.

11:24 Emma Sjöström: No. So this would just compare like, I

11:28 Amanda Mitt: Menu.

11:28 Emma Sjöström: annually except so if somebody else has already added it, for example, or

11:33 Amanda Mitt: Okay.

11:34 Emma Sjöström: Like this, and So we're just starting out kind of light but then we need to also consider Let's say you want to ingest like a lot of AR AP data, right at some point. How to avoid also. But I think it's a little bit of matter. What we need to Experiment with that data.

00:00 :

11:58 Amanda Mitt: Okay.

11:58 Emma Sjöström: And yeah. So would it be acceptable? Then it's my what I'm hearing that the current UI could support it for now because we also have a project internally for a kind of rematching, what all like booked actuals, how we would present them throughout the UI. How you will be able to like kind of view, the compositions of your forecasts, and all of those things.

12:25 Emma Sjöström: As well. So it's clear like

12:29 Amanda Mitt: Mm-hmm.

12:30 Emma Sjöström: If it's Machine learning prediction. It's this or if it's like AP or Yeah. Manually added payroll for example.

12:42 Amanda Mitt: Yeah, I think it's very clear. Yeah, you have the different colors. I think it's very pretty visual as well. When you see what's machine, what's Already, it's pretty visual. So I think it looks good and then if you go back, we had the description, right? So, we'll see. The category.

00:00 :

13:07 Emma Sjöström: But sorry now it just started over again. oh,

13:12 Amanda Mitt: No, I think.

13:12 Emma Sjöström: She would you would say Select that for now in the future our goal is to automatically also detect the category for you. But for now, we figured out, we'll Start using AI for the columns. And then you meant this one or

13:32 Amanda Mitt: I see still just, I don't know, maybe it's my Internet but I just see the forecast upload

13:38 Emma Sjöström: Oh no my screen, froze, fun. Okay, let me reset left.

13:42 Amanda Mitt: I think for as easy this way, right, Rodrigo, because we have like the different forecasts, we don't have like a mix of categories. Usually we have one with talent, that's payroll, one with tax. So should be pretty easy to Delete it.

13:57 Emma Sjöström: That sounds great.

13:59 Amanda Mitt: Does the open air open AP? Then would be a guess. Yeah, thinking about that file. Rodrigo. Do you have a sample quickly? Just so we can show them or maybe I can show Hex. Maybe you have it as well. I'm just thinking about how we would add the categories.

00:00 :

14:21 Rodrigo Cabrera: Yeah. Right. That one we've shared already but and it has a car. It has all the

14:23 Christian Sobkowski: We have the we have a file actually.

14:25 Rodrigo Cabrera: yeah.

14:26 Amanda Mitt: We have the budget.

14:27 Rodrigo Cabrera: And it has all the category. Yeah, budget and flow called you.

14:29 Amanda Mitt: Yeah. Okay.

14:31 Emma Sjöström: Yeah.

14:32 Amanda Mitt: Okay, then should be fine, I guess.

14:37 Emma Sjöström: We think so. So this is what I'm hearing something. That would like be usable for you. Even in this like fast version of the future.

14:48 Amanda Mitt: Yeah.

14:50 Emma Sjöström: Christian.

14:53 Christian Sobkowski: Is it okay? If I ask a few questions ever?

14:55 Emma Sjöström: Yeah.

14:58 Christian Sobkowski: And how? How often do these in? Or let me ask differently these input files. So think about is not the hex files necessarily but think about the talent files for example. How? That say messy are they? So when you try to make sense of them like is there You find errors in them? Are they do? They always look the same? Are they always like populated the right way? And do these files change over the course of a year.

00:00 :

15:41 Rodrigo Cabrera: A paranormal. I don't think they've changed that often like the layout. There's a new tab I guess every Every year. So when are there new

15:55 Christian Sobkowski: Yeah.

15:58 Rodrigo Cabrera: The. Yeah, whenever there's a new year, there's a new file. I can quickly, share my screen if you want to see them.

16:09 Emma Sjöström: Sure.

16:16 Rodrigo Cabrera: so, this will be This is the most recent one and it's a little bit of fun like more cleaner than the others. The Candidate Payment Tracker. Eh, this is for for expenses, basically your reimburse reimbursement. But for example we have the tax payment tracker, it has it has all of the information usually in the correct places.

16:46 Rodrigo Cabrera: In some cases, they have the payment type if it's an actual payment or if it's a refund, the account that we should be paying the amounts. Third amount. Payables currency in the Date that. The payment was made. So, for example, all the ones that have empty, they it

17:06 Christian Sobkowski: that's,

17:10 Rodrigo Cabrera: means that they haven't been paid. All of the ones that already have a date here. It means that they were already paid. A. Yeah, the consultation they should be booked and we have the same for direct payments and indirect payments, like tax payments and the same logic applies right now we will see a lot of them already closed.

17:37 Rodrigo Cabrera: A Because the payment run runs today. But they're still a few that are open and or waiting to be paid in the same with talent. I think it's a little bit different layout or template but the same logic, all of the the information which entities paying a the amounts in the date that it should be.

18:03 Rodrigo Cabrera: A paid.

18:06 Emma Sjöström: Right? So those have these okay, just to understand this palette, I clearly see. It's expected payment dates in the future. Did the other file also have that or was it just like when it's actually been paid?

18:22 Rodrigo Cabrera: A. This one. Let's see.

18:30 Emma Sjöström: Thing.

18:31 Rodrigo Cabrera: Payment DIA has led the payment due date. Yeah.

18:36 Emma Sjöström: And you would expect the payment due date and to be kind of the forecast, like the forecasted flow that show.

18:41 Rodrigo Cabrera: Yeah. Having some cases, it's just ASAP that doesn't help much. eh, well yeah there's usually a due date on On the payment.

18:56 Emma Sjöström: Gotcha. But those ASAP items then that's all. So interesting. I think. How would, how would you imagine something like Is that something you would want in your forecast?

19:12 Rodrigo Cabrera: the I think if it's us a ASAP like because we also have the requested date 25th of April, for example,

19:21 Yulia Ershova: Kids when it is paid actually for this example, if it was 28th requested, When did they really pay?

19:32 Rodrigo Cabrera: A 25th of April, it was a Friday. If even if it is a ASAP, the payment runs are

19:33 Emma Sjöström: Yeah.

19:38 Rodrigo Cabrera: only on Thursdays. So it would have gotten paid the Next week on the first of May. I think, and if we check, Comment from. Maybe on the second of May. So it even if it's a because the payment runs are only once a week. It was the following Thursday.

00:00 :

20:08 Yulia Ershova: so basically, if the program can put it, for instance, let's say all the SF positions for the earliest available Thursday,

20:17 Rodrigo Cabrera: Yeah.

20:21 Christian Sobkowski: Was about to ask, how you how you would want to utilize this. It's Because you would. So in this case, so this this is a kind of messiness. I was referring to. I'm assuming, right? That there is, there's probably other things, when, when you go through file, where something is like a little off.

00:00 :

20:41 Rodrigo Cabrera: Okay.

20:42 Christian Sobkowski: Would you want to? Because in these payments you would still want to have them in your forecast, right? It's really crucial for some For you to get them into the forecast. Would you want to? Amend that. Let me ask the different question. Would you want to go into the tax payment sheet and add a date there? Or would you want to change the date in in palm for example when you when you import this? So right, we're importing 10K ASAP.

00:00 :

21:21 Emma Sjöström: So yeah.

21:22 Christian Sobkowski: That ASAP information needs to come from somewhere. Would you sheet or in the

21:24 Emma Sjöström: So, what?

21:26 Christian Sobkowski: up?

21:28 Emma Sjöström: How it would? Yeah.

21:29 Rodrigo Cabrera: Yeah, we can we cannot a Change the tax tracker so it will have to be on the second. Stage of it after we downloaded it or after we share it with respond.

21:45 Emma Sjöström: Yeah, just the current flow. Did this assumption that you could just have a new

21:45 Christian Sobkowski: Yeah.

21:50 Emma Sjöström: file of all the rows that didn't have complete data, would that be acceptable to like edit dot file? Would setting a date that you want to use in your forecast.

22:05 Rodrigo Cabrera: Yeah, I guess so. Yeah.

22:09 Amanda Mitt: And would be able to upload the raw kind of like this. I will we would have to have some kind of data. Cleaning. beforehand to Be a little bit more effective in the beginning.

22:23 Emma Sjöström: It's all active, right? The idea was to help you easily kind of distinguish between so you have the green ones that are all good to go and then potential duplicate kind of yellow ones.

22:35 Amanda Mitt: Okay.

22:35 Emma Sjöström: And then we read ones that perhaps they or has ASAP which is not a validate say, right? So then the idea was to kind of let you download those errors into an oil and then Amanda's as you see, you could just re-upload that file for example.

22:58 Amanda Mitt: Okay.

22:59 Christian Sobkowski: Was, was that any question Amanda? Or were you asking about whether you can take the tax file directly and upload it?

23:07 Amanda Mitt: Yeah. exactly like yeah, because I know Emma's file,

23:09 Christian Sobkowski: 11.

23:12 Rodrigo Cabrera: It's like a Google Drive link.

23:14 Amanda Mitt: Yeah.

23:15 Christian Sobkowski: Yeah.

23:15 Emma Sjöström: Yeah, no. Yeah, the idea is that you will be able to take that tax file that you

23:16 Christian Sobkowski: Yeah.

23:20 Emma Sjöström: have for sure. And for the SO the LLM or AI would map the columns you would need to maybe first time, just double check the mapping inside full columns but yeah for sure that

23:34 Christian Sobkowski: The.

23:34 Emma Sjöström: that's the idea.

23:37 Amanda Mitt: Cool. Thanks.

23:41 Emma Sjöström: Of course.

23:46 Christian Sobkowski: I have one more question. As, as we showed this to you, another customer. They actually started. Finding all kinds of other. Forecasts or forecasted items that they would like to upload where they have some visibility on. Would that be the same here or?

24:18 Amanda Mitt: I mean, I see maybe one not immediately, I think maybe could come up like with our fees for example. Some of the, for example, for from the rcf or something like this, but not immediately, I don't see as a first like it's not a priority, but maybe if we

24:34 Christian Sobkowski: but,

24:38 Rodrigo Cabrera: so,

24:38 Amanda Mitt: get very confident with it, I think it can start. But then we could just use the manual as well. Like, not a Google Sheet. I think ideally, we could just

24:47 Emma Sjöström: Yeah, if they're just one ops. I think that's maybe easier. But that being said to Christian's question about being in the UI, that's definitely something we can work on moving forward and something we have some ideas around as well. If that's That's something you would prefer to maybe set a date.

25:07 Emma Sjöström: With the missing date for example that you could set it directly in the UI.

25:16 Amanda Mitt: You yeah.

25:17 Emma Sjöström: Yeah.

25:18 Amanda Mitt: If it's I think both works but that would be nice if you could just already fix it immediately, that would be cool.

25:28 Emma Sjöström: Gotcha.

25:31 Amanda Mitt: So when we still need some time rights before we start using it, or

25:39 Emma Sjöström: It's currently in development where? Yes. Hoping for one one, maybe two most prints. so,

25:48 Christian Sobkowski: period, doesn't

25:52 Emma Sjöström: Really get it out to you so that you can use it and then you can feed back on

25:52 Rodrigo Cabrera: Perfect.

25:56 Emma Sjöström: the usage and tell us what's working and what's like, you know, hey, not so

26:00 Christian Sobkowski: Seizure.

26:01 Emma Sjöström: good.

26:03 Rodrigo Cabrera: Yeah, I think. I have to drop for an article. I think Julia too.

26:09 Christian Sobkowski: Yeah.

26:10 Rodrigo Cabrera: but, Let us know anything Amanda.

26:16 Emma Sjöström: Well, do you have five minutes to find Rodrigo?

26:17 Rodrigo Cabrera: Thanks.

26:19 Yulia Ershova: See you, bye.

26:20 Emma Sjöström: To get quickly. So the settings on the account or should we take it another session, maybe?

26:25 Amanda Mitt: I have I have minutes if you want to. I'm free.

26:28 Christian Sobkowski: Emma. I I I'm gonna need to drop but you're free to go. Feel free to go.

26:30 Emma Sjöström: All right.

26:30 Amanda Mitt: Okay.

26:32 Emma Sjöström: Yeah. Do you have five? I can show you. Yeah.

26:35 Christian Sobkowski: Amanda Sue.

26:36 Amanda Mitt: I do, yeah.

26:37 Emma Sjöström: Okay, nice. So, just then in relation to this, right? So we will have the machine learning

26:40 Amanda Mitt: Thanks.

26:46 Emma Sjöström: predictions and we'll gradually build out. The booked actuals starting with this manual file uploads. We also want to explore, and this is a very simple, simple prototype. So, Imagine, this is our like, an account page with a full cost. It's Not looking exactly the same.

27:10 Amanda Mitt: Mm-hmm.

27:11 Emma Sjöström: You could somehow see like, for each of your category. And of how they're predicted in some way. So this is the simplest kind of Start. And then you could, perhaps imagine you would have something like forecast settings. Right now, it's at the account page, you could imagine, it could be a global setting too.

27:35 Emma Sjöström: Perhaps if you wanted to. To have that. So a bunch of categories, right? And initially, they're all

27:41 Amanda Mitt: Yeah. Like we need without intercompany that would be super helpful for example.

27:43 Emma Sjöström: machine that Yeah, exactly. And And for the purposes here, right? Let's say payroll now we don't want machine learning. but then it's a little bit about, like, What would instead like, what would be your options right now? This one default to zero. It's basically like no forecast, right? Or you want to, you know, manual you have a process where you manually monthly want to get this data in or in the future, that could be some integration or sometimes you just want to maybe, Set a fixed recurring amount.

00:00 :

28:29 Amanda Mitt: Yes.

28:29 Emma Sjöström: At a at a specific date. This is very basic just like let's say week like every month 20th.

28:35 Amanda Mitt: It's but it's a lot I remember in first noted on, but in plastic experiences like it would be. Yeah. Sometimes you just want to adjust the estimate kind of a scenario. So you had sometimes. I know it's not ideal, it's a bit dirty but it could help. Sometimes just Have a little bit more.

28:54 Amanda Mitt: I don't know, conservative view of the forecast or So in this case, I think it could help.

29:01 Emma Sjöström: Yeah.

29:02 Amanda Mitt: Yeah, that's very cool. I like how easy it is because in caliber, or in sap, whatever. It's always kind of very Yeah, it's a lot of work to Update all the food tests and like just yeah, you you always have to take it

29:17 Emma Sjöström: Culture.

29:20 Amanda Mitt: back to excel in the end. So I think it's nice that you can see all of it in the tool. I think this is a big. Yeah, it's very looks very easy to use. so,

29:31 Emma Sjöström: Got. And a quick question, we'll get well I'm sure we will talk more about this but Would it be valuable to have a global setting as well? So imagine you could set Payroll. will always be a fight, like a Manual input. For all accounts and then go in and adjust the specific accounts.

29:55 Emma Sjöström: Or would you feel that you'd rather go account by account? Anyways?

30:02 Amanda Mitt: For us global would make sense.

30:04 Emma Sjöström: Yeah.

30:05 Amanda Mitt: Mm-hmm. Yeah, I don't because we will have I can I can give you a deal list as well.

30:08 Emma Sjöström: Cool.

30:12 Amanda Mitt: Like for example, what would come from? M. So, for example, cachines, I think a lot of it from the DC. I think that's very like, maybe we can add a little Assumption as well. If for growth, I think that's the only kind of concern, but then for, yeah, I can give you like, we would need tax manual like manual for tax manual, for salaries.

30:31 Amanda Mitt: It's pretty simple, in our case, there's not. And for this kind of feast that we have sometimes

30:38 Emma Sjöström: Gotcha.

30:38 Amanda Mitt: And that but I think all of the rest and then open AI open. If you would be ERP data, plus Can we have? We can have both right? No yeah like

30:49 Emma Sjöström: Oh, that's a very good point if you want a combination.

30:52 Amanda Mitt: Yeah, combination because I think for opening our open AP would be a bit of a

30:54 Emma Sjöström: Nice.

30:57 Amanda Mitt: combination from machine learning and open. ER, and AP.

31:05 Emma Sjöström: That's really okay. So I'm I'm gonna I'm gonna let you go. That was super

31:05 Amanda Mitt: Yes.

31:09 Emma Sjöström: helpful just to get us started in. You know, making sure we're in the right direction. But I would love to talk more to you, obviously about these things. And especially I found the combination use case super interesting in terms of because we can support the combination That would love to talk more about how you want to see, like visually.

31:31 Emma Sjöström: See those forecasts, right? And how is that helpful to to make sense of it and what data is what and and all of that? To stop make sense.

31:41 Amanda Mitt: Yes.

31:42 Emma Sjöström: Yeah.

31:43 Amanda Mitt: Yeah, I'll continue asking the team as well. Thanks for showing me.

31:48 Emma Sjöström: Of course. Well, and nice seeing you always and thank you for staying over time.

31:52 Amanda Mitt: Right. No, I'm always happy to talk to you. It's always nice to see what's coming. And I'm, yeah, I feel like we're getting where we dreamed of. So thanks.

32:04 Emma Sjöström: That's that's really cool there. Okay, but let me know if you have maybe, you know, sometimes you get some idea afterwards or something like that. Just Just let me know.

32:17 Amanda Mitt: You too, bye.

32:18 Emma Sjöström: Thank you. See.