Inside Scale AI's finance tech stack and how they use TeamOhana

Kenny Tran: [00:00:00] Hi everyone. My name's Kenny. I'm the head of Business Transformation at Scale ai. I'm here today to talk about our journey with, uh, team Mohana, uh, fun fact. I have twin boys that are four, and I named them Greg Otto, Liam Franklin. So their initial spell golf. So I love golf. Um, if anybody's wondering, um.

Huh uh, like nine. That's good. I'm all right. I, I, I don't have enough time to play. Um, but our journey with Team Mohana, um, I think, uh, I'll get into it, but our whole goal was to make headcount planning a lot more simple. Um, when we first started, um, before my time, everything was in Excel. Um, may, who is here with us as well, um, used to run.

Headcount planning in G Sheets. Um, she would run it across multiple G sheets, um, that were all securely controlled, um, just so that a few [00:01:00] people had access, department leads, et cetera, hr, and a lot of her time would be spent consolidating all this data into a consolidated view, um, for headcount planning.

And it was a very painful process. Um, so I joined in January, 2023. Um, the first thing that my boss, Sandy, at the time, had asked us to solve was headcount planning because she didn't know how much longer may would be able to survive in this world of manually planning and excel, uh, in G Sheets. So. The, my first task was to look at our current landscape.

So at the time we had adaptive, we still have adaptive, um, we are migrating off adaptive, but at the time we had adaptive planning, right? And so most people ask, you could do somewhat of headcount planning in adaptive, which is true. What we typically lack in adaptive is the integrations, the availability of like greenhouse data, bamboo data, real time being able to understand.

Where a rec is in a certain process, getting recruiters visibility, [00:02:00] getting HR VP's visibility, getting all your hiring managers into adaptive is really unfeasible, in my opinion. Um, as much as you give 'em access, enable them, train them, they're not gonna log into the system. Like nobody ever logs into Adaptive except.

FP and a really, um, so that was like our whole genesis of like looking at different tools, right? So our first version of looking through and doing POCs on vendors, um, happened in March. Um, we looked at Headcount 365 as our first tool as well as double Fin. At the time we were doing a POC, um, we basically ended up with Headcount 365.

We thought the tool worked pretty well. Um, it would be able to ingest the data that we needed in terms of recs. Um, give all of our hiring managers access. Um, so we signed with 'em, we went through an implementation process, took about two months. Um, we had greenhouse integrations, we had bamboo integrations.

Um, and so we, we built everything. We went live in May. [00:03:00] Um, and so this is what I call it, perfect storm. Um, what had happened was Jenny, I boom happened in May. Um, so our business historically was mainly a VCV. Um, so a lot of labeling for cars, um, autonomous vehicles. And so we didn't really have a Gen I business until May.

Uh, and so when May happened, we went from maybe hiring like a few people a month to like 30, 40 a month. And so we started getting a bunch of requisition requests right off the bat once we went live. As we started going through there, we started realizing how painful it was to actually run through this in a system and that we didn't actually have scaled up to test.

And a lot of what we were planning on doing was, uh, kind of like improving it over time, but we just didn't have that capability, right? So we went through this whole process. People were still using it. Um, but the sad face is basically like by November we had department leads coming to our C ffo saying, it's taking us 15 minutes to log a [00:04:00] requisition.

We're, there's no way we're using this tool. Like you guys are gonna manually go and add it for us because we're not gonna do it. And so a lot of these struggles was around kind of like, if you think about comp bands. If I'm a, a hiring manager going in submitting a requisition and I put in a job role location, um, and a level, uh, and, and I'm assuming a comp band should return, right?

But we didn't have that in headcount 365 at the time. And so that's what a lot of the pain points that we ran into. And then when November came around, we also started planning again, right? So we do biannual planning, we do, uh, a half at a time. And so November came around and we needed to start planning, right?

So I was working with May. We're like, how do we actually plan out and build these scenarios of what, like an optimal headcount experience looks like? How do we incorporate. Um, a lot of our HR VPs, how do we incorporate our department leads, our leaders, et cetera. And so we asked s finally, they basically like, uh, we can't really support you right now.

So [00:05:00] we're like, okay, so we're basically gonna do this in G Sheets again. Um, so basically we ran the planning cycle and G Sheets again. We suffered through it. We went live, it was a struggle. Um, and then we basically knew at that time that we were gonna have to look for another vendor around the time, right?

So April comes, uh, a little bit before April comes around. We started doing a POC again on a lot of our vendors. So we looked at Team Mohana, we looked at Double Fin again, and we did like a whole comparison against existing roadmap of headcount. 3, 6, 5. The outcome of that. Basically after the first meeting I had with Tushar and another, I think one of the other individuals on the team was like, okay, this product actually can do a lot of what we want it to do from a planning cycles.

Um, the first thing that I was actually impressed with was like the scenarios, right? Like in a scenario you can actually. Have and create and see real time what the cost impact is. You can make swaps. You can make trades, and then all you have to do is click, merge, approve it, and it goes back into your plan.

Right? Like it makes it so easy and like we wish we had it for planning. So we went through, we signed with [00:06:00] them eventually. Long story short, we implemented them. Um, we've been super happy with the progress they've made. We've continually provide our feedback. We are very. Vocal customer, I would say, um, myself and May, we don't really sleep, so I'm up all night, um, just like submitting enhancements and like talking to thar, but the product's in a great spot.

We love it. Um, and it's, it's, it's been great. So that's kind of like our journey. Um, this is our process flow. So like what happens, like all of our hiring managers have access to Team Ohana. Um, so when they go in. It's to either create a request, a net new request, or a backflow request. So they go create a backflow request, they have somebody leaving on their team.

They go in, they click the individual that's leaving, and the data pre-populates in some scenarios. Um, or, and or if they're switching, kind of like the job roles or whatever, they can make those adjustments in there. It goes for approvals. Um, once it gets approved, recruiting leads get a notification, recruiting leads, then we will sign a recruiter.

A [00:07:00] recruiter can then initiate the job opening, uh, creation in greenhouse. Um, we go through the hiring process. Offers are sent and accepted. We have an integration through Workato that basically creates an inter, uh, that creates a message in Slack. Um, somebody in HR ops approves. It creates the record in Bamboo hr.

Bamboo HR eventually syncs back to Team Ohana and we have our reconciliation. So that's kind of like the overall process that runs today. There are one-off scenarios that we still are working through, um, typically like transfers. Um, we're still working through that. We get transfers. We have to kind of handle those manually today, but we're still trying to figure out a better way to do that.

Um, but this is our overall tech stack as well. So on the finance side, team, Mohana, adaptive, moving to Fantastic. On the people side we use Greenhouse, bamboo, and Workato is our integration middle layer that runs anything that we can't run, uh, through, uh, like any native integrations, for example. Um, use cases.[00:08:00] 

So again, um, back to scenarios planning, H one, H two, um, for example, like, I think I was talking to May yesterday, she basically just created a scenario invited all of our chief of staffs department leads, leaders, executives into this version. Everybody has limited control, so you can only see what you actually have access to and everybody's gonna go in, make their requests, they're gonna run through that, see what that looks like, and whittle their way down.

Um, and then at some point we'll convert it into our plan. Okay. Um, they can also, we also use scenarios for incremental unlocks through the year. If somebody has a huge project that they're trying to get like 20 headcount for, they can run that scenario in there. Um, as well as swaps, trades, anybody who wants to up level, down level, whatever they want to do typically runs through a scenario.

Um, we've enabled our hrps, um, to run through scenarios as well, so they are able to go in, run through some type of org change and we kind of. Process those through the merge, um, visibility and approvals. Um, so 100% of our backfills [00:09:00] and incremental as runs through Team Mohana. All of our approvals are done where hr, VPs, finance, bps, uh, department leads all approve.

Um, we have full recruiting visibility, so recruiters have the visibility of what they need to recruit against. The data is synced to Greenhouse to help them. So historically they would have to go get a rec and then go and create the opening and the job manually within Greenhouse. Today that's pretty much all automated, um, and we remove a lot of the manual spreadsheet trackers from a finance forecasting perspective, the prior process.

It used to be hours of reconciliation. So we went through this process. I think in the, when we had headcount 365 in May was out on parental leave. We had to do a bunch of reconciliation. It was probably the worst time of my life, like, I don't know why I'm doing reconciliation, but I was doing reconciliation, um, with the other finance person and it was miserable.

I don't, it was, I don't ask me. Um, but basically, uh, we would do the manual reconciliation, load the data into adaptive [00:10:00] manually. Um, and then get off all of our data. Um, our current process is, uh, team Mohana sends us a data set. We take that data set and we load it into adaptive. The future of this is what we're thinking through is between Team Mohana data, both in our current data plus our scenario data.

If we can get that through an integration, we could pipe it to Snowflake. Snowflake runs into fantastic. We really have really live views. Within our planning tool. And then we'll build a toggle so that any finance person can go in and say, I wanna create this scenario. And this is what, this, what scenario that I saw in Team Ohana, and I wanna run the full cost so that we have it in Fantastic, is basically running, uh, like 401k cost, payroll, tax, et cetera.

All the stuff that is like gets you to like a fully loaded headcount cost. Um, but those are the use cases that we solve for. Um, so roughly calculated, ROI is 183%. We've saved 90% plus of time in headcount reconciliation, um, 85% efficiency [00:11:00] gain during the planning cycle, um, and 99% of alignment across the org including confidential roles.

So we do support confidential roles through team on as well. Um, and the time to submit a headcount request for our hiring managers is less than two minutes.

Um, the future. So I've talked a lot about kind of, we, we we're in ai, so we do talk a lot about agents as well and kinda like figuring out how agents can help us do our work. Um, so in my vision, um, kind of like trying to get Tohar to push this roadmap in the team, Oana is like the system of action. Um, so I think.

With whatever they're coming out with. I think the retrieval and analysis of data within Team Ohana with the security layer, I think is like table stakes that says like typical rag pulling data. That's very easy to do. Right. What I think we should try to get to, and I think this is a lot harder, is assist in building out these scenarios, right?

So if I want to create a list of open requisitions at the start of my year, I know I have 500 people I wanna hire. [00:12:00] How do I split this apart of like EPD, I want to cap it to a certain budget and then I wanna distribute it across job levels, hiring seasonality level, location, and kind of like look at my recruiting capacity and gimme like a first draft.

Then I can go look at it or finance can go look at it and determine what they want. Or the other example is, I have a current scenario and I wanna optimize this to reduce my cost by 5 million. How do I go do this? Gimme a recommendation. Right? So a lot of this is kind of what we envision as like the future of agents 

Guest 1: with the second one first.

Um, the optimize kind scenario. That's, 

Kenny Tran: that's, okay. Perfect. 

Guest 1: That's coming 

Kenny Tran: first. Okay. Perfect. So we can get it. And that's all. So, any questions?

Guest 1: Yeah, tell me a little bit more about. Fantastic. Is that adaptive or? 

Kenny Tran: [00:13:00] Yeah, it's like an adaptive replacement. It's a, it's a much smaller company. I know most people probably use like Pigment or like a lot, a lot of the other bigger companies, um, fantastic is a smaller company, but their underlying tech is pretty advanced where you can combine, um.

Like, like typically like Anaplan pigment, you have sparsity issues. Um, so you combine like a hundred dimensions together. You eventually run into some, a lot of unused space and you run into issues, um, with fantastic, they solve the sparsity issue. So you can combine a hundred dimensions together with values and typically you shouldn't, there shouldn't be no limit to that.

So it is able to calculate a lot faster and a lot more advanced models. But yeah. Have you got budget manage. Using it as well, not just, uh, we're currently, we're, we're finishing up implementation, but that's the end goal is to enable all of our users to get in there. Yeah. Yeah. 

Guest 1: Quick question. So you've gone through a cycle of like nominal growth and then a lot more growth.

Mm-hmm. 

Kenny Tran: You were 

Guest 1: using sheets before, now you've [00:14:00] got the stool. Do you think this tool is, there's a use case sort of for both growth phases? Not every company is gonna be hired. 40 people. 

Kenny Tran: Sure. Absolutely. I, I see it both ways, right? I, I think regardless of whether or not you're growing a lot of headcount, I think it still helps in terms of like the overall reconciliation, being able to see live, like what your headcount looks like, giving visibility to hiring managers so they can see their org, they can see kinda like if I'm like a VP and I have multiple orgs, I own multiple cost centers, I can see how the breakdown happens.

I can kind of like do my planning, um, if you will. Um, and then I can also make the adjustments and work with my HR VPs. 

Guest 1: Got it. Yeah, that's helpful. 'cause I think that the selling point, or typically, um, talking to our CFOs was like, we're not hiring a ton. Mm-hmm. Why do you need a tool? Sure. I mean, headcount plan, this is beyond just hiring Right.

Obviously. Um, but it is making that case, so mm-hmm. You know, sort of how you would make that case depending [00:15:00] on the, this amount of health that's happening. 

Kenny Tran: Yeah, I mean, yeah, I, I think outside of that, right, I think you have the integrations to help recruiting. So a lot of that is between, within the rec ops and the recruiting optimization of being able to, um, have the data in Team Ohana go and push to greenhouse to an opening.

So like if you change or up level. It'll also do that for you. So you reduce a lot of the kind of overall maintenance of what a recruiter typically has to do. So that's like one use case. And I would say there's probably a few others in there, but I think the finance piece also helps as well. Still being able to track your headcount and being able to get that to your planning tool and running your cost analysis.

Yeah. Yes. Um, Hillary, I, um, Chris, I have a question about how you work on, I guess, change management. Sure.

Yeah. Yeah. Actually adopting a tool, especially at like the VP leadership level? [00:16:00] Yeah. Um, good question. I would say when we transitioned from sheets to headcount 365, that was a lot more rougher, especially with the overall experience that VPs and other individuals had going into the tool. I think. As we moved to Team Ohana, we made a deliberate approach of making sure there was a lot of training and enablement tools to kind of like enable our users to understand how simple the tool actually is and that we're not trying to overcomplicate it.

And then in terms of the push for like planning and getting VPs and everybody looking at the planning cycle. Into like the one scenario. I think that a lot of that was made and kinda like understanding, uh, or being able to push the chief of staffs and everybody to say like, Hey, this is our one single consolidated view.

This is all, this is. Where all you have to do is go and put in the same as right all you're doing is putting your job roles, your levels, your locations, and you're gonna get a comp number that spits out from there. And then from there we can build your plan and see what your overall like rough.

Estimated cost is 'cause we do like a 20% [00:17:00] overhead. Uh, but it's not like exact numbers, but it's close enough where we can like start the conversation and then work with our finance business partners to adjust down from there. But I think that's kind of like the approach we took. Um, does that make sense?

Guest 2: Um, I'm curious to learn more about the evolution of your role. 

Kenny Tran: Mm-hmm. Unique title. Yeah. 

Guest 2: Maybe what. What advice you give to people and companies that 

Kenny Tran: don't have the advocate transformation? Sure. Um, so I guess my role is more or less a typical business systems or business application role. So I was hired under finance, I center accounting now, um, but more or less, um, and regardless of where I sit, I, my team supports go-to-market systems, finance and accounting systems, and we do all integrations, automations, and now we're building agents.

Um, my team more or less. Is is pretty cross-functional already. I think the transformation part of what we do is kind [00:18:00] of like, we also kind of influence the process more than just like doing the application work. Um, but I guess to your question on like advice or kinda like who don't have the org, um, that's a hard que um, I would say try to find.

I, I would say so the way that we think about it too, and I, I was like, just like to step back is like we do a lot of build versus buy decisions and our build versus buy decisions is that we run very lean as a team. On my team today, we only have. Three other individuals in the US and we have a few offshore folks, but because we run so lean, we have to optimize for what will create tech debt.

And so when we can find a vendor that can build and kind of focus on developing the tools that we need without increasing our tech debt, that's kinda like the vendors that we like to, I like ideally purchase because we know that there. Their tool or their tooling will grow over time [00:19:00] without needing for us to do so much maintenance work, if that makes sense.

And so the way I would position it is that if you can get a tool that generates a ton of value for you with very little kind of like legwork or maintenance work, then that's kind of like the value in the transformation.

Any other questions?

Guest 2: One question. 

Guest 1: I know, I'm thinking, how do you know if a vendor is gonna, um,

Kenny Tran: that's a great question. Uh, that's a, that's a fantastic question. Um, so typically, um, uh, so, so. No, it's definitely not an easy question. So, so the way I think about it is typically when you engage with a vendor, you know, like when they respond, how they respond and like how fast they respond. [00:20:00] Um, I think for us, where it made sense that Team Mohan worked was like I could literally ping char at any moment in the day.

And I would get a response, um, and like, not, not just him, but like the rest of the team. Um, so just being able to understand that and like say like if I had like an enhancement request, right? They're not taking it lightly, but they'll like consider it. They'll talk me through it, we'll talk through it together.

Kinda like things like that as like the engagement that we look for. Like we have vendors where we know that if we started a. Like a dialogue with them or like we log a support ticket, right? We're gonna get stuck in the support queue that would never actually get a human to listen to an actual request.

And I think that's where we do a lot of our, when we do a lot of our evals, we run through that to understand how the vendor operates. And typically, if you also. I could say like the bigger the customer, you uh, bigger the vendor, you obviously have less impact on their roadmap versus like we are, I know we came into it as a pretty big customer for Team Ohana and they are a smaller company, so we have some influence, um, or a lot of influence over their [00:21:00] roadmap so we can actually drive what we want.

And that's kinda like the dynamic that we kind of try to balance, especially when the market is much more nascent and we don't have like a bunch of big customers, uh, or big vendors that we can look at. Um, and I know for sure like, uh, back to like our, like we use Bamboo, right? Like most companies at our size won't use Bamboo.

Um, but we've made it work through like our integrations, through our boomerang integrations. We, we do a lot of like integrations to make bamboo work for us, but ideally, we'd eventually move to Workday, right? But nobody wants to commit. To move to Workday at any point in time in the near future. So we're stuck on bamboo and, and like, it kind of just works, um, and we make it work.

But I, I, I think it, it really just depends on, kinda like the interaction with the vendor, um, is kind of how we come down to the evals typically. Um, but it's definitely like a lot harder to to, to tease that out, especially during the evaluation process. So 

Guest 2: another question on like, when you actually decided on like Ohana and [00:22:00] implementing the tool.

Did you guys kind of stick with like, let's start with one org team first and like test how this goes? Or like, were there any changes in terms of implementation and pivot up where you wanna start it, where you want it to start? 

Kenny Tran: Uh, no. I think when we, uh, we, we had buy-in from, uh, rec ops, finance, uh, HRBP, so we had buy-in from pretty much everyone to move off headcount 365.

And for us it was a very easy, like, we're just gonna do everything all at once. We're gonna make it work. Um, we're gonna build any integrations that we need to that is not natively supported. We're gonna do whatever it takes to get us to, to like a successful implementation. And that's kind of how it worked.

Like for example,

many.

Like planning, like outside, like from like a [00:23:00] finance perspective? Yeah, so we have adaptive, we're moving off adaptive to Fantastic. And that's where, where we run, we will run all of our, like our like payroll taxes on any other kind of like totally fully loaded costs plus vendor expenses, et cetera. Yeah.

Cool. Alright. Thank you so much.

‍