Why Finance Should Borrow From Engineering

Charlie Liu (00:02)

Hello everyone. Welcome back to Making Cents with Charlie. Our guest today is a co-founder and managing partner of SkySight Analytics, a boutique advisory firm specializing in finance and data readiness for mid-market PE-backed companies. He's a CPA, CMA, and co-author of Who Moved My Data? He's one of the first CPAs in Canada to serve as a chief data officer and has led finance functions through three successful exits over 20-plus years. Straight from Bali: welcome to my show, my mentor and friend, Greg Hood. Greg, good to have you. Your background looks gorgeous.

Greg Hood (00:40)

It is, it is. Bali afterwards. And no, I will say, you call me your mentor, I also call you my mentor. You are so successful, beyond me now. So I don't know who's a mentor anymore.

Charlie Liu (00:52)

I wouldn't say that, but I'm quite grateful for that, Greg. I think some of the best relationships are the ones where you learn from each other, right? Just now we were catching up behind the scenes and immediately we were nerding out as well. And today I think we're gonna have the same thing once again. Greg, we'd like to start with a rapid-fire, just to get us nice and loose, have a little bit of fun, and then we'll get into some of the fun stuff, which we'll nerd out on afterwards. First one: whiskey or wine?

Greg Hood (01:26)

So, whiskey when I need to think, wine when I figure it out.

Charlie Liu (01:32)

Ooh, I like that answer. Island or jungle? I feel like I know this one, but...

Greg Hood (01:38)

Yeah. Jungle. You've got the jungle up...

Charlie Liu (01:40)

[laughs]

Greg Hood (01:41)

...right here.

Charlie Liu (01:45)

Tableau or Power BI?

Greg Hood (01:48)

Honestly, the best BI tool is the one your team actually uses.

Charlie Liu (01:53)

I like that one. It's not shelfware if people are using it.

Greg Hood (01:56)

Yeah. I've used both Power BI and Tableau. It just matters whichever one your team's gonna use. That's the best one.

Charlie Liu (02:04)

I'm interested in this one: your funniest corporate jargon that you've used or heard.

Greg Hood (02:11)

So, the funniest corporate term... well, so I...

Charlie Liu (02:11)

Or hate. Yeah.

Greg Hood (02:14)

...actually was consulting with a Catholic-based, faith-based hospital system. So we were not allowed to use any violent terminology. So such things as, we couldn't say "bullet points," we had to say "dot points." We couldn't "execute" an agreement, we had to make sure we signed the agreement. So everything we said had to be nonviolent.

Charlie Liu (02:38)

I had a client, this was many years ago, I forgot the name of it, and I said "two birds, one stone," and the whole room froze. They were like, "We can't use that language here." So that was a similar case as well.

Greg Hood (02:57)

Like, no. We couldn't try to hit the target. We just had to hit the goal, right? So, like, the...

Charlie Liu (03:01)

Oh my god.

Greg Hood (03:02)

...there is so... you don't realize how much is in your vocabulary. So it took us, you know, a couple...

Charlie Liu (03:06)

Yeah.

Greg Hood (03:07)

...of weeks to get ourselves straight. And we actually had sticky notes on the side of our monitors to remind us: don't say this, say this, don't say this, say this.

Charlie Liu (03:16)

What was the favourite one?

Greg Hood (03:18)

It took me forever to get rid of "bullet points" and say "dot points." I just could not stop saying that.

Charlie Liu (03:22)

Yeah, I was gonna say, that's a hard one. Paper book, e-book, or audiobook?

Greg Hood (03:31)

E-books all the way. Yeah, I read 20, 25 books a year. Plus I travel a lot. So trying to lug around 25 paperback books would be really hard. And I've just never gone to the audio stuff. I like the e-books.

Charlie Liu (03:45)

I'm trying to get better with reading as well. I don't know if I have ADD or something, but I found a format that works now: I have the physical book and I have the audiobook at the same time. And that double consumption finally works for me, where I can sit down and actually go through a whole... well, not maybe a whole book at once, but sit down for three, four hours and actually focus on reading something. Speaking about books, Greg, you released a book, was it two years ago?

Greg Hood (04:20)

A year and... yeah, about a year and a half to two years ago.

Charlie Liu (04:22)

A year and a half ago. Who Moved My Data? How did that come to be? I know you did it with a co-founder or co-mentor. Mentor, yeah.

Greg Hood (04:30)

Co-author, yes. So it was with Craig Fookes. He was the first CFO I worked with. So back then I was an AVP of reporting and analysis. He's the one that actually taught me my first data warehouse. So in some aspects he was my mentor. So when he retired, we wrote this book together, part of it as his legacy, part of it also as trying to inspire the next generation of finance geeks: why you need to understand data. Because in my opinion, finance and data are joined at the hip. And as everything evolves, really, as a finance professional, you really, really need to understand data in the next generation.

Charlie Liu (05:08)

Completely aligned. I think a lot of folks come to me for guidance, especially in finance. A lot of accountants. And I often tell them accounting is a great start, it's really important. It's the language we speak. But for you to do well in finance, I think it's far less accounting, actually, and far more data and far more business understanding. And actually, I think a new one now is systems thinking, where you're thinking about the whole flow, which I think mattered less in the past, because now AI can fill in the gaps for a lot of things. But for you to build something that's reusable, scalable, you actually have to think about processes and data movements and the whole system.

Greg Hood (05:54)

Absolutely. I've worked in companies where we had 18 different systems feeding into finance, right? And that was primary systems, even secondary or tertiary systems. That's where some of the stuff at SkySight actually came to fruition: understanding that you have to think A to B to C to D to E, and how do you make that whole entire daisy wheel chain work? So system syncing is exactly where you need to be at. And systems... the foundation of all systems is data. So how do you get that data to flow from here to here without actually having to go like this, right?

Charlie Liu (06:31)

I wanted to dig into the data and AI, because I think that's where everyone's jamming and talking about these days. But I wanna kinda wrap up a point about your book. What was the aha moment that triggered the book, or what was your favorite piece of the book that you love to share?

Greg Hood (06:49)

A favorite piece of the book I love to share is just talking about the results that came out of the book. And not just that, the book talks about, hey, we went from, no, 10 business days to 9 a.m. business close, 90% revenues booked on business day one. That's kind of some of the aha moments. But really it's when I start talking to... I do a lot of work with university students, and I work with the CPA Society doing mentorship for our first- and second-year university students. Seeing the look on their faces when I did my first speech, that was a homemade aha moment, and I said, "Hey, you know what? I should probably write a book about this." And so now I have the book to substantiate all the speaking engagements I do with the students.

Charlie Liu (07:32)

That's a great combo. I do apologize, I haven't had time. I just wrote it down. It is a to-do that I'm gonna check in for my PTO, for vacation...

Greg Hood (07:42)

Mm-hmm.

Charlie Liu (07:43)

...next week. I need something to do while I try to disconnect from work. But AI, data, I mean, that's a big thing that everyone's focused on. The cost of inaction is higher than ever. I know you've been geeking out on it quite a bit recently. Your ally of choice is Claude, right?

Greg Hood (08:04)

Yes, it is, currently. Yep. I say "currently" because a year and...

Charlie Liu (08:07)

What is the...

Greg Hood (08:08)

...a half ago it was ChatGPT, right? So it's ever evolving.

Charlie Liu (08:13)

Yes, I recently switched over to Claude, I wanna say March. It hurts me, because I was a kinda early investor in OpenAI and I felt like I was just cheating on my spouse, switching over.

Greg Hood (08:23)

Mm.

Charlie Liu (08:23)

But I am loving it so far. Before the podcast started, we were talking about scale and AI, and there was a line that you mentioned to me last time when we caught up, which is: it feels like what VBA was back 20 years ago. And I think about that line, maybe not daily, Greg, but weekly. What did you mean by that?

Greg Hood (08:50)

So what I meant by that is, AI is a great tool, and Claude's a great tool. Claude Desktop's a great tool. But, you know, I was having a conversation with a senior accountant just about a week and a half ago, a great skill. And he was talking to me, he's like, "I've got this great skill, I'm gonna go sell it." I'm like, "Well, first thing you wanna do is check your employment agreement, because anything you build at work is actually the IP of the company, not yours. So don't get yourself in trouble from an employment agreement standpoint." But B, I asked, "Did your controller know about it?" He said, "No." "So what would happen if you leave the company right now?" He goes, "Nobody would have access to my Claude Desktop." So what happens to that skill? It goes away. It goes with you. So now you become a single point of failure. Well, this is no different. And I was that guy 15, 18 years ago. I wrote these great Excel macros, and I left the company. Well, nobody knew how to write that macro. It broke. It took them two months to re-close their books, because no one knew how to write the macro except for me. So all we've done now, we've got all these senior accountants writing all these skills in Claude Cowork and Claude Desktop, but no one's actually transferring the knowledge upwards. We've just accelerated the single point of failure again. This is a rewrite of VBA and Excel macros from 20 years ago. We're just rewriting history, but we're just calling it something else.

Charlie Liu (10:14)

And I think it's even more unique now, because you can transfer all the MD files, all the documentation, but I think for most people the setup isn't reproducible, because a lot of it's based off the local memory that I have for your account. And it feels to me that what AI is doing now is making a huge distinction: if you're using AI, you're just gonna get that much better, and if you're not, you're kind of being left behind. But where I'm going with this is, it doesn't feel very collaborative. It's now making super soldiers, just making you that much better. And I've spent a lot of time figuring out... I'm hearing the term "AI OS" or "company OS" these days. I'm trying to test it, and I haven't found an answer yet. But have you seen any work, or anything that you've learned recently, on how the infrastructure should be set up, how the documentation and the MD files should be stood up so that it's reusable by others? It's scalable. It's not just you doing your job better. Because the other thing I find is that AI is a very personal thing. The skills and these preferences you have, it's very personal. What have you seen? Yeah.

Greg Hood (11:34)

Yeah. Well, first thing, I will say that the engineering teams have done a way better, much better job than the finance teams have, because they automatically default into thinking about enterprise. So they all have the enterprise accounts, the corporate accounts. Well, I'm seeing on the finance side, everyone's just still working on their free accounts, or they have the Pro accounts, the $20 accounts, right? So you don't have that enterprise mindset. Everyone's just trying to close their month-end faster and make my life easier. So the best ones I'm seeing are the ones that, hey, we're actually thinking about it from the enterprise, they can steal it from engineering teams. Because they know the engineering teams are actually still using GitHub. They're still transferring all their knowledge back into a centralized world. The finance teams aren't doing that. Some of the ones I've seen are really great. They actually started putting together documentation, and putting that documentation, even of their prompts, into a centralized SharePoint or their version of GitHub, just to be able to say, "Hey, here's how I actually accomplished that." And I was with one client, and we're talking about, hey, your monthly team meeting: make sure everyone presents at your team meeting what you did in AI or in Claude, or, you know, a lot of companies are still using our friends over at Microsoft. So whatever tool you're using...

Charlie Liu (12:47)

Yes, just compiling. Yes. Yeah.

Greg Hood (12:50)

...present it on a monthly basis as part of your team so everyone understands it. But then you need to talk to your engineering team, make sure that gets rolled out, talk to your legal team, make sure you aren't sharing the wrong knowledge, right? Your AI that you're doing on your desktop has to be enterprise-ready.

Charlie Liu (13:07)

You brought up GitHub. Because I'm not an engineer, we're not engineers. When I started testing around things, we felt we were able to build a bunch of really useful skills for ourselves. You've seen it. The senior accountants you spoke about, they're seeing it. But we struggled trying to recreate the skills and the infrastructure for the rest of the team. We tested heavily with Notion. We actually tried using Notion almost like GitHub, and we didn't really understand what GitHub was, because again, finance accountants. It started working. Instead of writing local reference files, and this is getting very technical for the audience, instead of writing local MD files, we're basically using pages in Notion as the reference point. It worked, but then the token usage exploded. We actually had to stop and rethink. I actually got a Mac just to test this better, because I think a lot of the builders are the engineers, they're on Macs. They're the ones dogfooding it all the time. And we've seen improvements just testing AI and Claude on Macs as well. But when we went back to just your local MD files, things got much faster. The token usage window, easily 10, 15, 20 times. And we've learned, I think, part of it is that now we're able to become a lot more technical, a lot faster, a lot easier. Which means there are a lot of engineering best practices, like GitHub, like Git, that we need to pick up if you want to get good with AI, if you want to create sustainable, long-term, scalable infrastructure.

Greg Hood (15:00)

As, you know, my chief data officer hat kicks in every so often, from the data perspective...

Charlie Liu (15:06)

Yeah.

Greg Hood (15:06)

...we always had some of the best practices in engineering. We had some sprints and all that good stuff, but we always documented every single thing we did. Every SQL query we've ever written, every Python query we ever wrote, was shared with the team, right? Finance just has to become that during all those practices. It's just not something that finance people are aware of. We've always loved spreadsheets. Every finance person has always worked in Excel. We're still going to work in Excel, but you still need to be able to now document what the heck you're doing, because Claude can write your Excel spreadsheet now, even if it takes five different lines and a formula, and the formula is this long, and you don't know the queries, and what the heck is this function? So you've gotta be careful with that, right? But it does give you that advantage, but we actually need to be more sharing rather than being siloed. And that's the kind of duality of AI, of everything: as you said, it creates super soldiers, we can be so much more productive now. But it doesn't actually help the big picture, because it creates that single point of failure again.

Charlie Liu (16:10)

Yes. And beyond just that, what I find is a lot of folks, because it's so fast now, aren't testing and reading the plans, the contents. So recently I actually reached out to a bunch of engineer friends to learn about how they think, how they build, because to me it comes in handy in terms of how to get better with AI. A good engineer spends a lot of time actually planning the build. Not just planning, but figuring out what is the litmus test, what is a good output, what is not a good output, what does the QA look like before they actually start building. I think that was the biggest opener for me. Instead of just building, because I think a lot of times they're like, "I'm just gonna go build a six-tab model and then figure out the design sometimes on the fly." But it's slowing down, taking time to figure out what to build, and that QA thing. That was the biggest breakthrough for me on my end.

Greg Hood (17:11)

Absolutely. Now, once again, talking about my CDO roles, one of my very first data warehouses, we locked ourselves in a reading room for three days, whiteboarding what we were building. Whenever we had a deployment, we had a QA person and we had a standard script. That's all the stuff we stole from engineering. So once again, finance needs to steal from data. Finance needs to steal from engineering. You need to plan what you're gonna build. I was as guilty as the next person. I first went into Claude and started building, I did prompt, prompt, prompt, prompt. And I got the 17...

Charlie Liu (17:41)

Yeah.

Greg Hood (17:42)

...tab spreadsheet that made absolutely no sense. It didn't have any good flows. I had hard-coded numbers over here and I had assumptions over here. I'm like, no, no, no, no. Wiped it from scratch, sat down for like an hour. What am I actually gonna build here? Thought it out, put all my thoughts on paper, and then actually started building it. That's the way you're supposed to do it. Because if you just give brain spew into Claude, garbage in, garbage out. It's just gonna spew garbage back to you, right? So you need to actually have some logic. I can try to understand you, but the better prompting and the better model you have in your head, the better the output's gonna be.

Charlie Liu (18:16)

Yeah, a hundred percent. Recently, the first project I did when I got into Claude, I think this was two months ago now, I wanted to clone a piece of me that's useful in the sales process for what we do in FP&A and FP&A systems: what are the service offerings, what are the four walls of what we do and what we don't do. Because I was so used to working with GPT, which is designed, I think, completely different, where it's not designed for engineers. It's designed for the mass consumers. You can just sit there and prompt all day. So I took the same habits I had, moved them over. I think the first iteration, after three, four days of me just chain prompting, it created something like 135 MD files, which I looked at four days later and I was like, "I have no idea what I built. I have no idea what's in this." I had to absolutely slow down, basically design it on paper, think through what I wanted. Learn that Claude has a planning mode or thinking mode, especially if you work with Claude Code. For anyone listening, it's not as scary. It's basically Claude Cowork, that's basically it. It can do everything it can. You don't have to be an engineer. It's not that technical, but it has a planning mode where you sit there, back and forth, and just plan what you're gonna build first before you actually build it. And taking that step makes a huge difference.

Greg Hood (19:46)

Yeah. I had one client send me a file, they were so happy with what they produced. And I went, "Okay, great, you're learning Excel, you're using Claude." And I went to the Assumptions tab...

Charlie Liu (19:55)

Yeah.

Greg Hood (19:57)

...changed a couple of the assumptions, and the output didn't change. There was no tie-in. So, back to that...

Charlie Liu (20:02)

Yeah.

Greg Hood (20:03)

...you still have to understand what you're building, right? And another one sent me... we were talking, and, dude, they started prompting right on the screen for me, and he started building this out. I'm like, "You've just got a mess right now. Do you actually know what this tab is doing?" He goes, "No, Claude just put that in there." Well, if you don't know what that tab's doing, you still need to understand it. Look at the function, he doesn't understand what this function is. He goes, "I've never seen that function in my life." Well, now if it breaks, how do you fix it?

Charlie Liu (20:34)

Yes. I'm seeing a lot of... who was it? I can't remember the company, but they released some numbers. This is a software company, I think it was Mozilla actually. They had a record high in releases, but then bug releases was also a record high in Q1 as well. A lot of it driven not by Claude, and I'm seeing the same thing in finance now. I'm seeing a lot more bot, models, where I'm like, "This is better than what I typically expect from this person six, nine months ago." But you dig into it, there's a lot of additional tabs, there's a lot of sections that aren't fully connected. People aren't thinking through the whole thing. What I found with our team that's been really instrumental recently is, before we start building, we write the test scripts, almost like you and I, as good consultants: what are we actually building here? What are the test scripts? And instead of jumping directly into the build, and everything we were talking about a few minutes ago, we write the test scripts, and AI is great at building those functional test scripts. Then we can actually iterate and look at this with the client, or look at it ourselves: are these the levers that we care about? Do they actually reflect your day-to-day business functions? And then we start building it. The test cases get us grounded from a functional, non-technical standpoint. It confirms this is what we want. And then we can use them after we build. And every time we make an iteration to the model, we go through the full test script again to make sure that it's QA'd, it's tested properly. A big piece now is, it's just so fast to build, right? It's actually the guardrails around it that you have to spend more time on.

Greg Hood (22:27)

Yeah. We've been talking about some of the negative aspects of it, but don't get me wrong, I love it. It is absolutely, like, we're getting steroids here. It's a superpower. I'm doing work in three hours that used to take me three days, right? It is a great, great tool. It's making us efficient. It's just making sure that you use it effectively and responsibly.

Charlie Liu (22:51)

Yes, and we're trying to figure out what we're doing with the superpowers, well, lightning in a bottle kind of thing. We've actually walked away from a lot of the fractional FP&A work we did. And what's really interesting is, a year ago we would find a replacement, someone else in our network. We've actually had recent clients where we sat down and gave them some of the Claude skills we built, gave them the guardrails, and some of them are able to maintain the model, actually even adjust the models themselves. Now with us around, there are risks, they understand it. They're choosing to try to own this themselves first before trying to switch into another fractional partner. But it's changing quite a bit in terms of how we work with clients. Now, Greg, I know for what you do at SkySight Analytics, you see a lot of different use cases. Where have you... what are some of the interesting use cases for AI? Or how has your work shifted in the last year, two years?

Greg Hood (24:03)

One thing that's made our life a ton easier is the data cleansing world. A lot of the stuff we've done is building finance data warehouses, especially in the M&A world. You get a lot of, hey, I've got three different databases. In one database you have "New York," and the other database has "NY," another database has "N.Y.," right? We used to have to try to build translation tables and go back in and filter and actually just try to cleanse all that data, and that'd be a full day's worth of work. Well, now we just have a skill in Claude that can run it and cleanse it for us in 10, 15 minutes. Of course, you have to test the output, right? So then we load it into our database and say, "Hey, you only have three distinct names now, and here are the numbers. And yes, the total still ties to the total," right? So stuff like that has made our life a lot easier. One of my co-founders... I couldn't imagine doing this work now without Claude. It's just changed our life, like, night and day on that type of stuff. And one of the really interesting things we're seeing now is, as you're getting ready for M&A work, and as a founder you're trying to sell your company, you're going to go through data room prep. We're actually starting to see intelligent data rooms now. So you can actually start to query the data that's in a data room. One of the big things we always said before, and it's something we do, is you have to make sure that your data is clean to go in the data room. We had one client, it's a great success story. They went from 3 million to 7 million to 14 to 21.6 million in revenue, as per QuickBooks. As per the CRM, they had 21.3 million in revenue. And as per the databases, they had 21.15 million. So now you have three different revenue numbers that the buy side is trying to look for, right? That's just disjointed data. So we came in, rolled out a finance data warehouse, reconciled it, and the number was actually 21.52 million. We actually added $2 million back to the valuation. But now you have the data, the reports all sliced and diced five, six different ways, that I can sit in the data room. You can actually query the data room and say, "Which file was it that showed me my revenue by salesperson again?" Or "Which was my utilization?" You can actually start querying it, which is really, really cool. I would have never thought about that two years ago: actually querying data rooms during an M&A process.

Charlie Liu (26:15)

Dumb it down, I haven't heard that term before: intelligent data rooms. Are you giving clients access to that data room and they're using their own agents or whatever to query the data? Or are you querying live? What do you mean by that?

Greg Hood (26:31)

So it's a third-party tool. We don't create the data room. The data room's a third-party SaaS tool that we partner with. So we just populate the data room and all the files with the data from our finance data warehouses, and then it takes care of organizing all the data and says who now has access to what: HR data, finance data, sales data. But then obviously there's just a chat tool within it. Any good piece of software now has a chat overlay, right? But, you know, because even a transition I did just over a year ago, the due diligence list was 306 lines long, which meant that we had about 300 to 400 different documents floating in a data room. Well, trying to keep track of 400 documents and remember what document was that? And of course, a transaction takes not one month, not two months, three, four, five, six months. We have to start uploading...

Charlie Liu (27:19)

Yeah. Yes.

Greg Hood (27:21)

...files on a monthly basis. So then you start getting 400, 500, 600 files sitting there. How do you keep track of them all? It's a big, big ask. And trying to remember what data was in what document, this little query to it, the intelligent data room, allows you to just query what document had what in it.

Charlie Liu (27:38)

Okay, got it. No, that's fascinating. I remember the one I prepared, I think this was about a year ago now, where they come back to you with lists of these questions, right? 20, 30, 50 questions, and you just have to give the results back to them. But now you can query these live. Sometimes they can do it themselves as well.

Greg Hood (27:55)

Yeah. We were on a call with one of the Big Four, actually. It was a standard two, three hour kinda reviewing this report, reviewing this report. And they were asking us, you know, can we get this report, can we get this data? So we keep on talking, talking, talking, and about 15 minutes after the call, we're having the wrap-up and they say, "Okay, Greg, when do you think you can get the data to us?" We're like, "It was in your inbox 17 minutes ago." And they're like, "Mm. What?" We're like, "Yeah, we just queried it, and we already have the data set. So we just queried it and we created the report, and here you go." They were like, "Wow. Just wow." And that's easily three days' worth of work at best. And now we just did it live on the call, and it was ready within 17 minutes.

Charlie Liu (28:42)

Things are getting easier. Things are getting faster. Where do you think... what do you need to... I don't know if that's the right word. Where do you think your focus should be to increase your value in these kinds of projects, these M&A transactions? Because searching, kind of the grunt work, is getting a lot easier, right?

Greg Hood (29:09)

Searching definitely is. But there's still a lot of foundational work that's missing. A lot of companies are exiting, a lot of them are high-growth companies that still haven't gone through data foundations. Does my CRM data tie back to this dashboard? Does this dashboard tie back to my financial statements? I've been guilty of this myself in the past, saying, well...

Charlie Liu (29:34)

Yeah.

Greg Hood (29:34)

...well, I know it's this dashboard. It's always off by $100,000, just don't pay attention to it. But now, during due diligence, people are going to look at that dashboard, right? Or it's like, "Yeah, I always know that this product is always misclassified in this report." Well, you can't have that during due diligence. Because every single question mark is a way for the buyer to go discount, discount, discount.

Charlie Liu (29:58)

Yeah.

Greg Hood (29:59)

So you need to have... a lot of that is not even AI-based. That's just data foundational. You can use the AI to help cleanse it. But now you have to go back not just a day, a week, a month. You're going back 24 to 36 months. So, yes, you can use AI to clean up, but you still have to understand that you still need to do that work. And we're still seeing a lot of people who miss that element of it. Like, "It's good enough." Well, it's good enough for internal reporting. It's not good enough for a hundred million dollar transactions.

Charlie Liu (30:33)

Makes perfect sense. It can help guide you, it can do a lot of work for you, but I think someone still needs to sit on top of it, manage it. Get the clients to actually...

Greg Hood (30:47)

And, no...

Charlie Liu (30:47)

...eat their broccoli.

Greg Hood (30:49)

Yeah. One of the other clients we're working with, they had a great forecast. "Yeah, we used Claude to create this great forecast." It's like, "That's awesome." It's fake. You had a 12% revenue increase last year, you had 6% the year before that, and now you're gonna do 82% year-over-year revenue increase. Just because you can use Claude to put together a... and Excel is still garbage at times, right? You can make a spreadsheet say anything you want it to say. And Claude will do that for you now. So you can make it say anything you want it to say, faster and easier. But it's still just a spreadsheet. And a spreadsheet's just numbers on a spreadsheet. It's not real life.

Charlie Liu (31:32)

And with AI, it's even more powerful. I mean, it's deterministic. It's trying to just go towards the goals that you want, right? I have seen cases where it just changed the underlying data, because it doesn't know what's right or wrong. It knows these are the goals, the growth and targets you want. And so sometimes the growth parameters are reasonable, but to get where it needs to, I've seen cases where it actually wants to change the historical data. There's a lot of nuances that you have to be careful about. There's a lot more checks and ties and guardrails that you have to put in.

Greg Hood (32:08)

I was working with a pre-rev company, and they're like, "We stand by our forecast. Claude did an S-curve, so we have 100% faith that we can go from zero to three million in the next 24 months." You're selling $85-per-month subscriptions. Do you understand what your go-to-market is? Do you understand how many subscriptions you actually need to sell at $85 a month? And by the way, are you actually still aware that SaaS valuations have gone down? They're like...

Charlie Liu (32:42)

Are you actually seeing this, where clients just have that confidence because AI has kind of laid out the numbers for them and they haven't...

Greg Hood (32:50)

Yeah.

Charlie Liu (32:51)

...tried...

Greg Hood (32:53)

It's an S-curve. They don't know what an S-curve was, but it's an S-curve. So...

Charlie Liu (32:55)

Yeah. Yeah.

Greg Hood (32:56)

...it must be good, because Claude said this is one of the foundational aspects of stats. It has an F-table and Z-tables, and the correlation response is 85%. So, 85% against what, though? Claude is testing itself. It's like, "I'm 85% confident in my own garbage." So yes, we're actually seeing clients who are actually believing, because Claude said, "I used statistical analysis, I used a profit model, I used this model." Think about it for a second. You are a team of three and you think you can do 23 million dollars of sales and ARR in 24 months, $85-per-month subscriptions. And I just don't see it.

Charlie Liu (33:37)

I think that's the part where, more than ever, that common sense to ask, to check, to look at the whole business, is more important than ever. I think AI can cover all this, but most people aren't asking the right questions, and the whole piece is driven off the prompt that you have. And a lot of people are still really bad at prompting. I found, at least from my experience, the best users of AI are... because people often think younger folks are better with technology, they're usually early adopters, but I actually find that managers, execs, leaders are using it far better, because they've just had more experience managing people. Managing agents is not that different. But also asking the right questions.

Greg Hood (34:28)

Yeah. I remember early in my career, one of the things I was taught was, I don't have to know everything, I just have to know where to get everything. Who can get me the answer? So now, yes, now I'm managing a couple of agents instead of managing all the staff. And, you know, I do feel bad for some of the next generation, because some of the junior work is getting eliminated, right? But there is that false confidence. Even one of my friends got Claude to do their tax returns. And I took a quick look at it, I'm like, "By the way, your CRA classifications are all incorrect." So Claude told you the wrong classifications on your business taxes. So you can follow it if you want, but you're probably going to get audited and fined. So you can't have blind confidence. Like I said, a lot of great things about Claude, a lot of great, great things. We're using MCP all the time, it's great stuff, right? But you've gotta be careful on the output.

Charlie Liu (35:27)

And the thing is, it just looks... that's the word. It looks so high value, so trustworthy right now. It's easy to just skip the homework and then just believe it with blind faith.

Greg Hood (35:40)

Yeah. One of my friends was actually trying to run his stock portfolio off Claude. And it's like, "Great, great, great. Here's a great investment." And it tanked. And, well, here's why it tanked. "Yeah, my bad." People have... it's that kind of slick salesman. You're just personable, you talk with confidence, so they believe you, right? Once in my career, one of my first managing jobs, I was just joking around with the staff, and they came with a question, they had this problem. I'm like, "It's because of this, this, this, this." So, okay, they walked away. I'm like, "Get back here. I'm totally pulling your leg. That has nothing to do with this. I need you to critically think about this stuff." And that's what Claude is sometimes. It's that fast-talking salesman sometimes.

Charlie Liu (36:32)

Never heard that one. It's right. Claude, I think, is a bit more real with you. If you talk to ChatGPT, I think some of the other ones, they do a great job making you feel like you are the king. It's smooth, it's smooth. That's a great analogy: it's a fast-talking salesman for that interaction.

Greg Hood (36:57)

Yeah. Well, according to ChatGPT, I'm in great shape. I'm really, really nailing those weights. I'm its best client for personal training. I don't think so.

Charlie Liu (37:11)

I love some of the things it says. It just tells you, "Such a great idea. Good for you." It really pumps you up, and it feels good talking to it. I think that's a big piece in terms of how they're acquiring users. It feels good using these tools. It's not just for the output, but for a lot of people, it's companionship. It feels good using it.

Greg Hood (37:38)

No, there are cases now of AI addiction. There are cases of people actually going into depression because of what AI told them, specifically ChatGPT, that wasn't really true. They weren't as great as they thought they were. They didn't actually realize that, no, they hadn't solved all the world's problems, and went into depression.

Charlie Liu (38:04)

Yeah, I mean, earlier you were talking about clients getting that blind-faith confidence. I've seen it with the folks I've hired as well. They very quickly fall apart after you start digging in and asking some critical questions, because you can tell they just haven't gone through it. They haven't sanity checked the guidance and the learnings they've gotten from it. There's a lot of blind faith. I think I saw a stat last year. For an average person who used to do research on Google, even with Google, which has done all the work for you, I think it was eight times, meaning every search you do, you go through about eight links to look around. That's all gone. Most people have zero... even for me, I'm guilty of this as well. A lot of times I'm not fact-checking. I'm just taking the answers directly as is.

Greg Hood (38:57)

No, no, just... they're great tools, you just have to use them efficiently, effectively, and responsibly. You still need that common sense to understand what it's looking at. It's great. Use it, speed you up. Just make sure that it's corporate-ready, enterprise-ready. Make sure that you actually test it out. You know, "Hey, I changed the assumptions page. Did the P&L page change? Oops, nope." Well, maybe we should do some testing, right? Like every movie already kind of spoke about.

Charlie Liu (39:27)

Greg, earlier you mentioned you feel sorry for the youth, the next generation, because a lot of that early grind work that built up the foundation for why you and I are where we are today is largely disappearing. For you, if you think about someone that's earlier in their career, or even for us, where we're trying to coach and wrap that next generation, things are quite different now, right? How would you think about how they should build their early phase of their career.

Greg Hood (39:59)

So, one thing, just to get on my soapbox here for a second, is I want to talk to the universities. A lot of it comes from the universities and colleges. They have to teach the right stuff. And some of the university academia is very, very slow at catching up to what's happening in the corporate world. You know, some of...

Charlie Liu (40:18)

Mm-hmm.

Greg Hood (40:19)

...the academia are like, "Da-da-da-da-da, it's still good, it's still good. You'll...

Charlie Liu (40:22)

Yes.

Greg Hood (40:22)

...get a job as a junior accountant, and you'll still be doing bank reconciliations."

Charlie Liu (40:24)

Yeah.

Greg Hood (40:25)

No. By the time this first-year class graduates, that first year is not gonna be doing bank reconciliations. And I should know, my very first job, I was doing 28 bank reconciliations every single day, up to 100 sometimes. That won't exist anymore, right? So I want to talk to academia and say, get your stuff in order and actually teach people useful skills. Update yourselves for AI. But beyond that, what I'm gonna say is that you, coming up in the next generation, have an actual wonderful opportunity. You can be AI native. Campfire has a...

Charlie Liu (41:01)

Yeah, yes, yes.

Greg Hood (41:02)

...brand-new... they're AI native. You know, NetSuite is trying to tag AI onto a 20-year-old system. I'm NetSuite now, right? The next generation is Campfire. They can actually come up being AI native. You can learn through it. That's the whole entire thing of, did I actually... no, yes, bank recs taught me a lot of valuable information. I would not want to do bank recs anymore, right? So if I can become that superpower and teach other people how to use AI, there's still a lot of people. I'm talking to a lot of SMBs and smaller companies, smaller ones, 10-, 12-, 13-person companies, who have zero usage of AI, absolutely zero usage. So there's still a very, very big opportunity for the next generation to come in here and teach people and be part of that change going forward.

Charlie Liu (41:59)

Adoption is always slow. Even most usage today is basically replacing their usage of Google now, right? So instead of searching for something, they're asking AI. That AI-first... I love your analogy. I need to ask myself if I'm Campfire or NetSuite, but that's something I have to sit alone with.

Greg Hood (42:18)

And so, sorry, my NetSuite friends. I am a user of NetSuite.

Charlie Liu (42:26)

Well, I mean, they have to be around for a while, right? It has to be a household name. But changes happen, right? You have to go with the times. But at one point I'll grab a whiskey and ask myself that question. But I think this AI-first question, it's a great opportunity. It's such an interesting time. But for someone younger, what does that mean? If you're able to give them practical advice, how do they get good? Especially, let's just take the context of what we do, finance, accounting, strategic finance, corporate finance, M&A, things like that. When they're coming out of school with very little experience, and they may not have the opportunity to go through the grind, go understand the foundation. How would you recommend them becoming AI-first? Is it taking that engineering, that build mindset and just building more? Is it thinking about and learning more of the business aspect? What is your practical advice for them?

Greg Hood (43:27)

Yeah, that is the crux of it all, right? It's easy to say "become AI-friendly, become AI-enabled," right? If I'm coming up, one thing I'm doing is I'm actually using Claude to teach me Claude. It teaches you how to use it, right? So you have to have that curiosity by nature. You have to have that drive. Because all of us at one point in time were newly grads, and we all have...

Charlie Liu (43:54)

Yeah.

Greg Hood (43:55)

...to kind of struggle. And the only thing we can ever say is, "What are you gonna bring to the table?" Well, I've got zero years' work experience. I've got, you know, an okay university average, GPA average, but I've got a really good work ethic, and I'm curious to learn, and I'm a sponge, right? That's what you have to bring. So take it upon yourself. Even my first job, I remember I would be working hours of overtime, unpaid, because I was listening to somebody else's problem over here and learning about, you know, fund...

Charlie Liu (44:19)

Mm-hmm.

Greg Hood (44:20)

...accounting. I came up through fund accounting and investments. And fund number 51 was having a problem. So I'd go and listen to fund 51, listen to the senior accountant talking, and them working through the problem. And I'd absorb that and realize, okay, well, that took me 45 minutes of my time. It's now 5:30 at night. I still have 45 minutes' worth of work to do. I'm spending the next hour at work just trying to learn, because I was over here. I've gotta catch up. But that's on my own. But you know how to do stuff like that. So I would be taking every single AI course I can get my hands on. There's so much stuff available. I'd be looking at what are the AI tools. We just mentioned Campfire, right? Another tool we've been working with is Nominal. I actually am a big fan of Nominal. They're AI-first, right? Start actually researching that. I'm here telling people this right now. So if any students are listening, research Campfire, research Rillet, another great one, right? Nominal. Take a look at these pieces of software. Maybe ask them, "Hey, can I get a half-hour course? Can you give me a demo?" Stuff like that. Learn that type of stuff. Because yes, business knowledge is acquired over years, but you can also learn and read about business knowledge. Talk to people like me. I'll give you 15, 20 minutes of my time. Ask me a question, I'll give you the answer, because it never hurts to have a conversation.

Charlie Liu (45:44)

I think that's the foundation for a lot of things. I think it's that curiosity, it's really, really important. And being comfortable asking for help. So all the folks listening, Greg and I, both of us, love to help. Just reach out. Happy to give you time. But that curiosity is really important. At least for me, what I've loved about AI is that it's so easy and so fast to learn now. And a lot of these platforms, like Google, spend a lot of their focus building studying tools, learning tools. I love NotebookLM. And one of the reasons I love it, I think two factors. One is it's really good at not hallucinating. The output that comes with NotebookLM is really driven off whatever you feed it. Now, if you feed it crap, that's a different story. But if you feed it good material, it's really just gonna come back with that material. And then the podcast function is phenomenal. It now has study guides and a bunch of other tools with it. But AI has made it so much easier to learn. But you need to have that curiosity, that willingness to learn. You can't just hear something, not understand it, and accept it as is. You should really try to figure it out. If I hear something I don't get, make it a goal where you need to understand everything you're seeing in terms of your line of work or your experience.

Greg Hood (47:11)

So one thing AI can't teach you is curiosity by nature. That's on us as humans.

Charlie Liu (47:18)

Do you think that's an innate kind of thing you're born with, or is it a skill that you can strengthen?

Greg Hood (47:24)

I absolutely believe you can strengthen it. In the nature versus nurture, I think...

Charlie Liu (47:31)

Yeah.

Greg Hood (47:31)

...in the right nurturing, anyone can learn that. You just have to be able to ask, you know, the five whys: why, why, why, why, why? One example is, when we were on my first data warehouse, we had a great sales quarter, right? "Why do we have this great sales quarter?" This fund went through the roof. Everyone says, "Yeah, yeah, great, great, great." But why did that fund go through the roof? BC was really great. Okay, BC had great sales. Great. But why? Well, it looks like Bob had a really great quarter. Why did Bob have a really great quarter? Bob just sold a lot of funds. So if you want to talk to Bob, Bob's like, "I tried this new technique, blah, blah, blah, blah, blah." There's your answer, right? That's the curiosity. You have to keep on digging into the why. So when you see something... I had this one report, this was one of my very first words of SQL. I was looking at this report. I had one number on page three. And then on page four, the number's off by one penny. Page five, same number as page three. Spent three days locked in a meeting room trying to figure this out. And then I became like the crazy mad scientist in the room with white papers...

Charlie Liu (48:40)

Yeah.

Greg Hood (48:40)

...all over, and somebody else joined me, and a second person joined me, and a third person joined me. And everyone's trying to figure it out, and I'm trying to make one-dollar entries into the system and it just won't correct. Well, someone else came by and said, "What are you guys doing?" He looked at us, walked back to his desk. He's like, "Try tonight." He spent about 10 minutes fixing something and it worked, and he goes, "This is History Reports. We actually have duplicate tables. This report was picking this table up, which rounds to four decimal places. This report picks up from this other table, which rounds to eight decimal places. That should fix it." If I hadn't spent those three days asking why does one report have a one-penny difference, I never would have learned SQL.

Charlie Liu (49:22)

It's crazy. You've shared that story a few times with me now. That's kind of triggered... your whole start with data and SQL really made who you are today, just asking those whys. And I think a big piece is that curiosity is often contagious as well. As people see you, they see your obsessiveness in terms of figuring this out, people want to help as well. I agree as well. I really do think it's something that can be built, right? So folks that are listening today, it is a must-have, I think, in terms of performing well, getting to the places that you want to be, but also especially surviving this AI apocalypse as well. I often work with my team, and I think the big piece I'm always trying to get across to them is, don't just accept things as they are, right? If you have a problem or you have something that you're trying to reach, don't just go to your manager and be like, "We have a problem." But try to get curious enough in terms of figuring out what are the options to get there. And then that's gonna start building that muscle to be curious, but also have that critical thinking. There's a frame of mind that I often recommend for the team. I think it's called 1-3-1, where: what is the one problem? What are the three options? What is the recommendation? And I think if you don't have curiosity, you're not gonna be able to go through that whole exercise. A lot of times you just expect someone else to spoon-feed you that answer, that recommendation. I love this piece about curiosity.

Greg Hood (50:59)

No, my very first manager role, I would not answer any questions about any of the mutual funds unless the juniors or intermediates came to me with a bunch of reports they'd written out, with a possible reason why they're having this problem with this fund. They'd just come to me and say, "Fund number 52 failed Rule 4," and expect me to teach them what it is. No, we have a book, we have documentation. Go try it yourself. Come back to me with a possible solution. Then we can work through it. But I'm not gonna give you all the answers. I want you to learn what Rule 4 is and why it failed.

Charlie Liu (51:39)

That makes a great, I think a great mentor as well. It's not our job to do everything for everyone. It's also not scalable. A big pivotal point I had in my career was actually working with my mental health coach, where I think for a lot of my career I always felt it was my place to take on all the weight and do everything for everyone. And he gave me a line which was really good, which is, by you doing the work, you're actually robbing someone else of the opportunity to learn and fail and go through this. And that made me that much better of a mentor and a manager. Because I think a lot of folks in the corporate world, they go from an extremely strong IC, and then one day you're told that you're now a manager, you're a people manager, and you have to stop doing the IC work. And that is really hard for a lot of people.

Greg Hood (52:35)

And, you know, the CPA Society teaches us accounting. The CFA Society teaches you valuations and risk. No one actually teaches you how to manage a team. I was actually lucky, I worked for a big corporation when I first started, and they sent me on a three-day HR management course when I got promoted into management. But a lot of people don't have that. "Hey, I'm great at month-end" doesn't make me a great controller who's gonna be able to coach a team of five people. Even having taken that course, my manager was asking me to do something. I said, "No problem." Manager asked me to do something, "No problem." The senior manager asked me to do something, "No problem." And he asked me to do something else. I'm like, "I can't do it anymore. I'm already working this weekend. I've already done 15 hours of overtime." He's like, "Greg, just because I ask you to do something doesn't mean you have to do it. You have five staff members. Use them." That was my... no. But because I was two months into being...

Charlie Liu (53:36)

Yeah.

Greg Hood (53:37)

...moving from IC to being delegate-down, right? That's a lesson everyone has to learn.

Charlie Liu (53:44)

Setting boundaries is also really important. Setting it up...

Greg Hood (53:46)

Very much so.

Charlie Liu (53:47)

...so that it's successful, so you set yourself up for success. Any kind of guidance that's worked for you in terms of managing others? Because a lot of people, it's a really difficult muscle in terms of how to communicate, how to manage, how to delegate. How do you think about delegation? What does that look like? How much input do you provide to the team?

Greg Hood (54:13)

That's kind of a loaded question. It all depends. If it's my senior, the senior gets a lot less guidance than the junior does, right?

Charlie Liu (54:16)

Yeah. Yeah. Yeah.

Greg Hood (54:21)

I also believe that as a CFO or VP of finance, wherever you happen to be, you still have the opportunity...

Charlie Liu (54:26)

Yeah.

Greg Hood (54:26)

...to train a junior accountant. I want to make sure I still have... if I have an opportunity, I'm gonna obviously talk to my manager or my controller and say, "Do you mind if I talk to a junior accountant, or work with this one person one-on-one?" And usually the answer is, "Yes, of course," right? No problem, right? And usually some of the juniors are like, "The boss is coming to talk to me, right?" But at the end of the day, someone taught me something, right? And I'm just a really big believer in paying it forward. So it's up to us to really train each other, the next generation, going forward. Someone gave me a break, so I want to give somebody else a break. But down to the delegation, it's tough at first. And it actually requires a bit of a spine. It requires you to be able to say, "A month ago I was your colleague, now I'm your boss." And so...

Charlie Liu (55:15)

Mm-hmm.

Greg Hood (55:15)

...you do need to understand that you're not sitting at this table anymore, you're sitting at this table now. And you just need to be able to have that confidence to actually say, "I need you to do this piece of work." And there's that referent power of being the boss, right? But you have to understand you cannot do everything as the boss. You actually need your team, and you're only as strong as your team. Because when I was at Canada Life, 50% of my performance review was based upon whether my team met their goals or not. So if I didn't make a strong team, I didn't get a good performance review.

Charlie Liu (55:52)

Are you more in the camp that you should be hands-off, or more in the camp that you need to be in the details with the team? Some people might call it micromanagement, some people call it leadership in the details. Which camp do you have a preference towards?

Greg Hood (56:14)

I fly at the 3,000-foot level and then do dive bombs into the 10-foot level when needed.

Charlie Liu (56:23)

I like that. For what we do at MCC, I've spent the last better year and a half... the big thing was basically coaching a lot of ICs to become managers and execs. And I made a mistake in the beginning of flying really, really high and being very hands-off. And then, as I learned over the last year and a half, at least for me, management in the details was really helpful in terms of helping them grow, but also pulling the company in the right direction. I think a lot of folks are too hands-off. You see this a lot of times with bigger corporations, where people in the company talk about how the folks at the top don't actually understand what exactly is happening on the front lines. At least for me, I find that you need to have a healthy balance of both sides. It's...

Greg Hood (57:24)

No...

Charlie Liu (57:24)

...important to get into the details as well.

Greg Hood (57:27)

Absolutely. Working with someone last year, they were going towards 100 million of revenue. They didn't really have a proper FP&A function set up. So I actually did the first 30 to 40% of the Excel model myself. Then handed it over. I did one department, said, "Okay, now hand over...

Charlie Liu (57:48)

Yeah.

Greg Hood (57:49)

...to the senior analyst," and said, "Now try to replicate the next four departments, what we did together." So there's that, you know, flying high, get down. Get down low. Now I'm gonna go back up and watch him try to do it, and then come back down as necessary. And then same thing: okay, we have an issue over here with this bank. Okay, I'm gonna look at the documents. Okay, now it's resolved. We're halfway there. I'm gonna start climbing back up again and start monitoring again. Because, no, good or bad, in the fintech world, the growth, the messy middle, the founder-led worlds, you don't actually have the opportunity to stay at 30,000 feet. If you're up there, you're just gonna fail. You actually have to be a hands-on operator. But you have to be... and I've never quite figured this one out: are you strategically operational or operationally strategic? But anyways, you gotta find that balance of being an operator and being strategic. I think that exists in...

Charlie Liu (58:38)

I think that's one of the things I love about you.

Greg Hood (58:40)

...and that exists in any company that's less than 100 million.

Charlie Liu (58:44)

I love the fact that you're also able to do both sides as well. A lot of folks can't. I think today it's more important than ever. I think there's a lot of people talking about the middle management, or pure people managers. A lot of that is disappearing. I think a lot in the Valley they talk about this a lot: middle management layers disappearing. You have to be able to actually understand the work. You have to be able to understand what exactly is on the front line. You're able to actually get down, roll your sleeves up, and solve and build the things that you need to, if needed as well.

Greg Hood (59:29)

Yeah. I remember talking to one company, and I was talking to the AVP, I think it was a director. They...

Charlie Liu (59:36)

Yeah.

Greg Hood (59:37)

...literally, four or five hours a day, they were just forwarding emails. Their job was they basically forwarded emails to everybody all day, because they wanted every email to come into them and they were the central hub. And that's not really a job anymore. You're just forwarding emails for four hours a day. That's just not an efficient use of resources.

Charlie Liu (1:00:00)

One of my team members, Peter Lay, who's one of the early founders with me, was spending a decent amount of time at one of the big tech companies. I don't think he wants me to say which one. But he was telling me every month they'd spend 15, 20 days preparing for that presentation meeting to explain what they're actually doing, instead of actually doing the work themselves. And it's refreshing, actually. I think a lot of that is disappearing. It's a lot of blow that a lot of companies have had for a long time. Those are disappearing. People are recognizing that this is not how business should be run. But also, to me, from a personal, humane level, I don't know how happy you can be if you spend most of your career just justifying what you do instead of actually doing the work.

Greg Hood (1:00:47)

I was involved in a debate last week. Should a board deck be 57 pages, or should the board deck be 15 pages?

Charlie Liu (1:01:00)

And what's your answer to that question?

Greg Hood (1:01:03)

Fifteen. Fifteen pages. You can put a bunch of stuff in the appendix, but if you have to have 57 pages at the board level, the board's not being strategic enough.

Charlie Liu (1:01:14)

Yes, and asking the right questions.

Greg Hood (1:01:14)

But, you know, because... yeah. And I was guilty. I figured the more information you get the board, the better the deck. Properly formatted, brilliant, brilliant effort, two weeks of hard work just to get this beautiful deck out, and then of course page three, page 17, and page 34 have different numbers on it. Like 21.5, 21.6, 21.5. Man. And by the way, Claude still does that. I actually caught a report last week where Claude had rounding issues. They had pages three, 17, and 24 all with slightly different numbers on them, even though I told it to double-check its work. It's like, "You're right. Great catch. You're the best. Thank you for catching that." Don't make that mistake in the first place.

Charlie Liu (1:01:54)

Yeah, yes, exactly. I think that's a really good tip just for anyone using AI. At the end of it, start a new conversation and just ask it to double-check the work once more. That has saved my ass multiple times. Just doing that one final QA on the work.

Greg Hood (1:02:17)

All my Excel models that I have Claude help me build now, and this is just good financial modeling practice, have all the checks and balances, a checks and balances page. There's this tab. Claude does not do that by nature. You have to tell it to do that, right? So I now have 22 different checks and balances on this to know that my model actually ties back across all the different views.

Charlie Liu (1:02:45)

Greg, I think this is a good point to end. We've had a lot of fun talking about AI this past year, giving guidance to younger folks in terms of how to figure out the career, the growth plan in this crazy AI world that we live in today. I know there's been a lot of advice you've already given out, but we do like to end every session with your two cents advice. What's your two cents advice, Greg?

Greg Hood (1:03:13)

That it's never over until it's over.

Charlie Liu (1:03:20)

I like that. I have a similar one where I go, "I didn't come this far to come this far."

Greg Hood (1:03:28)

Yeah. The story behind that was that it was one of my dad's favorite sayings. And he taught us that when we were eight years old. We were losing six-nothing with, you know, 10 minutes left of the game: it's never over until it's over. Which actually, Yogi Berra was the one who said it to begin with. But it stuck with him, and it stuck with me my whole entire life, that, you know, it's never over till it's over. You can always fight to the end. And if you get knocked down, you get back up, because it's never over till it's over.

Charlie Liu (1:04:00)

I know we're coming to an end, and I actually think that's a big piece for some of the younger audience: that resilience, that grit. I don't know if it's more important than curiosity, but it's up there. You have to struggle, and that's okay. It's part of the journey. It's part of the fun as well. But as Greg says, it's not over until it's over. Well, thank you so much, Greg. This was an absolute master class in finance tech, maybe even a bit of career planning as well. And for everyone who's listening, thank you so much. Thank you.

Greg Hood (1:04:37)

It's a blast. I always love talking to you, Charlie.

Charlie Liu (1:04:42)

Same, Greg. I love it. Thank you.

Greg Hood (1:04:46)

Peace out, everybody.

Charlie Liu (1:04:50)

Awesome, dude.