Skip to main content

Larridin raises $30M to continue to help enterprise organizations understand the impact of their AI investment   Read the Press Release      Watch the Video

See our AI Impact podcast episodes on YouTube.

In the latest episode of AI Impact, Dan Zhang, CFO of ClickUp, joins Larridin co-founder and CEO Russ Fradin. She shares the frameworks she uses to evaluate AI vendors, measure ROI across a finance team, and convert skeptics into believers without hype or slide decks. Dan oversees finance, accounting, legal, HR and operations at ClickUp, a global productivity platform with more than 1,000 employees, and she explains how she thinks about AI investment, adoption, and the new hiring profiles she's building her team around.

Key points

Dan told ClickUp's CEO when she joined five years ago that her job was to enable growth and get the company ready to go public when the time comes. Her approach to AI balances the same two goals. Several key points emerged.

A 2x2 for deciding where AI money goes

To sort AI vendors, Dan borrows a 2x2 from a pricing expert: how autonomous a tool is, and how much value you can attribute to it. The holy grail is high autonomy and high attribution, such as customer support agents that close tickets on their own; either the ticket got closed or it didn't. Low-autonomy, high-attribution tools, like coding assistants and AI notetakers, are helpful, but their value is diffuse, so she monitors them closely and budgets for them as a share of payroll, perhaps 10%.

High-autonomy, low-attribution spending, like platform and security investments, is part of the overall OpEx portfolio. Everything else goes in the sandbox: small budgets, around $100 per tool and about $100,000 across the business, for people to experiment. When a sandbox tool shows it belongs in another quadrant, its champion comes to Dan for an investment conversation.

Three kinds of AI ROI, and the platform comes first

In finance, Dan sees three kinds of AI ROI. The first is straightforward automation, such as journal entries, accruals and reporting. The second is new capability: at a product-led company with a long tail of small invoices, AI can now review every one of them. "Don't boil the ocean" used to be the rule; now you can. The third is a quality boost, from C-minus to A-plus.

The two-to-three-day business synthesis she used to write as a chief of staff now arrives automatically from ClickUp Brain. Before buying a new tool, her team's first question is whether an existing platform already does it or soon will. When a promising AI accounts-payable startup came along, it turned out Coupa was launching the same thing the next month.

She expects automation to consolidate onto existing platforms, citing a Morgan Stanley survey in which CIOs ranked application software first for consolidation, while new vendors win on new capabilities and quality. And she tailors how she talks about AI impact: time saved for frontline employees, OKRs for leaders, and lasting business-model improvement for the board.

Move the middle with proof, not hype

Dan sees three groups in any company: the AI-forward, the resisters, and a large, capable, curious middle that doesn't know where to start. When an analyst on her team built an AI agent on top of a semantic layer for financial data, his colleagues met it with skepticism.

At her HR partner's suggestion, Dan held an in-person offsite where the analysts tried to break the agent with edge cases. By the end, the mood had shifted from "AI can't do my job" to "I can test a hypothesis in seconds instead of hours." Dan's broader point is that people, process and systems, the domains of HR, the CFO and the CIO, can no longer operate in silos if AI is going to deliver productivity gains.

Russ, who sold software to much of the Fortune 2000 at his previous company, agrees: anyone who thinks AI will take over organizations overnight has never had to get people to use new software.

Hiring for judgment

For Dan, AI changes the hiring question from how many people you need to what's the best use of human judgment. She's looking for two new profiles: context optimizers, who keep data and processes clean, consistent and AI-ready, and insights interrogators, who challenge AI's output and own the final judgment.

She evaluates global talent on quality rather than cost, and expects AI to rewrite long-standing "corporate math," such as account executive quotas and sales engineer ratios, as AI takes over admin work. Above all, she hires for curiosity and humility. Candidates may use AI in interviews, but they have to show their prompts and how they iterate.

Conclusion

See the full episode on the AI Impact Podcast playlist on the Larridin YouTube channel. To get a demo of Larridin, contact us.

Quotes

  • "We always like to say, 'Don't boil the ocean.' But now with AI, guess what? You can boil the ocean now." (Dan Zhang)
  • "Your goal is not to use AI. Your goal is to grow the business by 50%." (Dan Zhang)
  • "This is not a financial decision. This is a quality decision." (Dan Zhang)
  • "We're doing an unusual thing now, where we're trying to retrain the entire workforce of knowledge workers globally." (Russ Fradin)
  • "I think the people who believe AI is just going to do everything in an organization tomorrow, and take everyone's job, have never actually dealt with change management and getting humans to use software." (Russ Fradin)
  • "The winning companies are going to say, 'If my employees can do four times as much, I want four times more employees.'" (Russ Fradin)

Q&As

Russ had questions; Dan had answers.

Q: How did you think about AI vendors? Was it "Let a thousand flowers bloom," a steering committee, or everything locked down at first?

A: I like to borrow a framework from a pricing expert. It's a 2x2. ... The two axes are autonomy and attribution: how autonomous an AI tool is, and how much value you can attribute to it. The holy grail is high autonomy, high attribution. ... And the last one is questionable autonomy and questionable attribution. ... I call that the sandbox. We still give a very limited budget there, for example $100 to play with. ... Once they see a tool can move into one of the other three quadrants, they come to me, and we can have an investment conversation.

Q: How much are you seeing the impact of AI on the work your FP&A and accounting teams do?

A: For AI ROI on the finance team, I always like to put it into three buckets. The first one I call one-to-10 automation. ... The second bucket I call creating new capability for the finance team. ... We always like to say, "Don't boil the ocean." But now with AI, guess what? You can boil the ocean now. ... And the third ... I call it the C-minus-to-A-plus quality boost. ... The work existed before, but the quality AI can provide is totally game-changing.

Q: If your FP&A team has a new request, is it "go vibe code it yourself," or "let's check out this new vendor"? Do you worry about vendor creep?

A: I think the challenge now isn't which AI tool we should use. It's that we have way too many AI options. So my first question to my team is always: great use case, but does our existing platform already provide that, or do they have it on the roadmap? ... I was shown a demo of an AI accounts payable solution from an early-stage startup. ... And the team said, "The Coupa team just showed us that they're going to launch it next month." ... So I said, just turn it on, and see if you get the same ROI.

Q: How do you think about HR's role in change management and AI in your company?

A: AI cannot move the needle unless you can move the middle. ... You have the AI-forward people, and you have the resisters. ... But you also have this massive middle. These people are capable and curious; they're just unsure where to start. ... At the offsite, I asked all the human analysts to play the role of challenging the AI. ... Gradually, I started to sense the shift in the energy in the room, from "I don't think AI can do my job" to "I can test my hypothesis in seconds instead of hours."

Q: How do you make sure you understand what "great in AI" looks like at ClickUp?

A: When you think about and talk about AI impact, it sometimes requires CFOs to use different languages. ... When you're talking to frontline employees: "What used to take me 30 minutes now takes me 30 seconds." ... When I talk to leaders, you need to attach the AI impact to their OKRs. ... Your goal is not to use AI. Your goal is to grow the business by 50%. ... And there's a third audience: the board. ... What the board is looking for is sustainable, long-term improvement in your business model.

Q: Has AI impacted the way you think about hiring?

A: It's not how many people do you need anymore. It's what is the best use of human judgment. ... I'm specifically looking for two types of people. The first I call the context optimizer. They make sure that the data structures and business processes are clean, simplified and AI-ready. ... The second profile I'm looking for is the insights interrogator. They don't just take AI output and accept it. They're critical thinkers.

Q: When you're hiring younger folks, is what you're looking for today different than five or six years ago?

A: I think the fundamental traits are the same: curiosity and humility. ... If AI gives you an answer, do you just take it and run with it, or do you say, "Hey, what about this angle?" ... So now, when I give cases to the people I interview, the first sentence is, "Yes, you can use AI, but show me your prompt, and let's do a live session together."

Topics in this episode

Key topics in this episode include:

  1. AI impact: Dan tailors how she talks about AI impact to each audience: time saved for frontline employees, OKRs for leaders, and long-term business-model improvement for the board.
  2. AI tool usage: Dan describes three groups of AI users at ClickUp: the AI-forward, the resisters, and a capable, curious middle, and how a live offsite where analysts challenged an AI agent won over skeptics.
  3. AI costs: Dan allocates AI spending with a 2x2 of autonomy and attribution, including roughly 10% of payroll for copilot-style tools and about $100,000 a year in small sandbox budgets for unproven tools.
  4. AI ROI: Dan sorts AI ROI in finance into three buckets: one-to-10 automation, new capabilities such as reviewing every small invoice, and a "C-minus to A-plus" quality boost.
  5. AI workflows: Dan expects AI workflow automation to consolidate onto existing platforms, while new capabilities and quality boosts may still come from new vendors.
  6. Human productivity with AI: After the offsite, Dan's analysts went from doubting AI to testing hypotheses in seconds instead of hours, and she says AI should let enterprise sellers sell about 40% more by freeing them from admin work.
  7. AI agent productivity: Dan says the holy grail is high-autonomy, high-attribution AI agents, such as customer support agents that close tickets on their own, and that AI SDRs and demo agents are already rewriting sales-team ratios.
  8. Enterprise AI: Citing a Morgan Stanley CIO survey that ranked application software first for consolidation, Dan predicts companies will work with fewer AI vendors over time.
  9. AI and the Board: Dan says the board isn't interested in quick wins like using AI for accruals; it wants to see sustainable, long-term improvement in the business model.

Full Transcript

See this latest episode of AI Impact.

Chapter 1: Preview and intro (0:00)

Russ: If you had to predict the future, do you think three years from today you're working with more AI vendors than you are today, or fewer?

Dan: I think that there will be a trend of consolidation on the existing platforms.

Russ: How do you make sure you understand what "great in AI" looks like at ClickUp?

Dan: You have a problem, and a tool to help you solve it in a much better way.

Russ: Has it impacted the way you think about hiring?

Dan: I think it forces us to answer a very important question. It's not, "How many people do you need?" anymore. It's, "What is the best use of human judgment?"

Russ: When you're hiring younger folks, do you find what you're looking for today is different than what you were looking for five or six years ago?

Dan: The fundamental traits are the same, and those are the traits that will determine if this person can shine in the age of AI.

Chapter 2: Dan's role as CFO at ClickUp: enabling growth and public readiness (0:51)

Russ: Hi, this is Russ Fradin. Welcome back. I'm here today with Dan Zhang, CFO of ClickUp. Thank you for joining.

Dan: Thanks for having me.

Russ: So I really want to get started with very basic things. We can talk about your career and things like that, but how do you define what a CFO really does? What do you think of as your real job?

Dan: I joined ClickUp five years ago. When I joined, I told the CEO that you hired me to do two things. The first one is to enable growth, and the second one is to make sure that the company can be public-ready when we have the scale. So I would say being a CFO, especially a growth-oriented, strategic CFO, is the constant optimization and balance between the two. You need to be the one who's enabling growth and helping the business hit the highest potential it has, while at the same time making sure that the business can be ready for the public market someday.

Russ: Really partnering with, I assume, the rest of the executive team.

Dan: Yeah, exactly.

Chapter 3: Managing finance, legal, HR, and operations under one roof (1:49)

Russ: And what are the parts that fall under the CFO inside ClickUp?

Dan: I oversee the traditional office of the CFO: finance and accounting. I also have legal, HR and operations. We consolidated all those functions under the office of the CFO.

Russ: Exciting. And obviously we'll talk a lot about AI. But I'm interested in how you think about... so how many employees are at ClickUp?

Dan: Over a thousand employees globally.

Chapter 4: Running a global, remote-first team of 1,000+ employees (2:12)

Russ: Wow. And mostly remote and distributed?

Dan: Yes. The majority of them.

Russ: Does being a global company cause a lot more challenges on the finance side? I assume so, but I've never really talked to a CFO about this. I assume it's much more complicated.

Dan: It is, but it's all about the ROI. We went global with our talent strategy because there was upside. This all-remote, global strategy opened up the talent market very significantly for ClickUp. If you look at our top performers, a big portion of them are from international markets. And of course, as a CFO also overseeing HR and legal, we need to make sure we have the compliance and the system hygiene to make sure all these global employees are being supported in the right way, and that we're following all the local compliance requirements. So the upside is big, and yes, it adds work for the back-office team. That's why you need to constantly keep an eye on how fast you're scaling, and keep the balance there.

Russ: I know remote companies seem to fall into two camps. Some are just heavily asynchronous. Everything's written, people get back to you, there's good respect for everyone's time, and it's a written culture. Others try to keep some standard set of global hours. How do you actually handle the culture-building part? What time do you have calls?

Dan: We have a hybrid culture as well. Because ClickUp is the all-in-one productivity platform, as you can imagine, we do a lot of async work. If it can be a strategy memo, why do you need a three-hour meeting? It has definitely improved efficiency, and it makes internal processes very ready for global employees to consume and get enabled. But we've also realized the value of in-person connection. So recently we started to roll out more standardized global hub offices, to cover as many employees globally as possible. They're all encouraged to go into the office, one day a week in some locations, two or three days a week in others, so that we can still build that camaraderie and some of the serendipity that only happens in person.

Russ: It's lonely. There's a lot of upside, but it's lonely sitting at home.

Dan: Yeah.

Chapter 5: How AI permeated ClickUp: top-down alignment meets bottom-up adoption (4:30)

Russ: So obviously I know most people want to hear us talk about AI, and I have a million technology questions. When it was clear that AI was going to be a thing... because you were at ClickUp before that, if you joined five years ago. How did it start to permeate the organization? Was it bottom-up? Was it developers? Was it your CEO saying, "We must do this"? How do you think about it structurally?

Dan: It starts with curiosity at the top. We saw it top-down: the executive team put our heads together and determined that this is the big trend; this is the future. Once we had alignment, we said we need to cascade that message down to the entire company: AI is unavoidable. This is where the future is going. The entire business needs to transform itself to be AI-ready. At the same time, we saw a bottom-up brush fire start to grow organically. You had people vibe coding at midnight and starting to build their own applications to improve productivity, and we realized that tapping into that bottom-up, organic energy would help us accelerate the whole implementation and adoption curve in the company. So I would say it's top-down met with a lot of bottom-up enthusiasm, and that's the only way you can roll out AI successfully in a company.

Chapter 6: The autonomy-attribution 2x2 framework for evaluating AI vendors (5:51)

Russ: And how did you think about vendors? Was it "let a thousand flowers bloom"? Was it, "We have a steering committee"? "We have a person in charge of AI"? "Everything's locked down at first"? How did you think about that? I know people don't really have a coherent plan for navigating it.

Dan: I think about this a lot, and I like to borrow a framework from a pricing expert. It's a 2x2.

Russ: Of course. It's always a 2x2.

Dan: Yeah. The two axes are autonomy and attribution: how autonomous an AI tool is, and how much value you can attribute to it. The holy grail is high autonomy, high attribution.

Russ: Okay.

Dan: Think of vendors like Decagon or Sierra. They help support close tickets autonomously. You can quantify how much value...

Russ: They either happily closed the ticket or didn't.

Dan: Exactly. So that's the holy grail. Then you have low autonomy, high attribution... sorry, things like code-gen products, Cursor, and some tools like AI notetakers. They're very helpful, but it's very hard to attribute the value.

Russ: Sure. We all know there's some value in having AI notes of all of our sales calls, but it's diffuse. It's not clear how to measure it, but we know it's valuable.

Dan: Exactly. How much productivity are these users getting? Yes, it's easier for them to write code, easier for them to record a meeting, but it's very hard to attribute that.

Russ: Did you sell more? I don't know.

Dan: Totally. So that area gets heavy monitoring. To your point, I already have a certain investment in payroll, in headcount. How much of that can I allocate? Maybe 10% of total payroll for these copilot tools for employees. Then there's high autonomy, low attribution. I call that quadrant the platform investment: some of the security investment you put in, for example. That's also part of your entire OpEx portfolio: how much budget you're willing to allocate there. And the last one is actually what you were asking about: questionable autonomy and questionable attribution. Should we just shut it down, or still let people play with it? How much money?

Chapter 7: The $100 sandbox budget for unproven AI tools (8:19)

Russ: Sandbox.

Dan: Yeah. I call that the sandbox. We still give a very limited budget there, for example $100 to play with. It's not going to break the bank. Across the business, I basically put $100K into an experimental budget like that. Once they see a tool can move into one of the other three quadrants, they come to me, and we can have an investment conversation. So that's how we think about the distribution.

Russ: So departments are enabled, within certain limits, to pick tools, and then they come to the CFO, the CIO, some version of that, if we really want to roll something out everywhere.

Dan: Yeah. And the framework helps me ask the question: great, shiny demo, but where would you put it? In the holy grail? Is this high autonomy but low attribution? Or is it a sandbox? Usually that helps push clarity: this is just a small spend, or we should get ready to scale it.

Chapter 8: Three-bucket ROI framework for AI in finance: automation, new capability, quality boost (9:15)

Russ: I have a bunch of organizational questions, but I hadn't really thought about it within the finance team. How much are you seeing the impact of AI on the work your FP&A team does, your accounting team does? How do you think about that, just on pure core finance work?

Dan: For AI ROI on the finance team, I always like to put it into three buckets.

Russ: Okay.

Dan: The first one I call one-to-10 automation. The very straightforward use cases: automating manual journal entries, helping you do accruals, helping with some reporting. Very clear. The second bucket I call creating new capability for the finance team. I'll give you one example. Because we're a product-led growth company, the volume of customers and invoices that we need to collect is very much higher.

Russ: Exactly. A long tail.

Dan: Previously, it was very hard to touch every user, to review every small invoice, like a $50 invoice. We always like to say, "Don't boil the ocean." But now with AI, guess what? You can boil the ocean now. So we've gained a lot of financial capability, because now AI can do all the bottom-tranche touching and investigation, and a lot of scrutiny there. And the third capability is the one I enjoy the most. I call it the C-minus-to-A-plus quality boost, in the areas that help my work the most. I'll give you one example. I used to be chief of staff to an executive team member, 10 years ago. My job was basically putting together a synthesis of what's going on in the business. When he came back from travel, I'd say, "Hey, here's what's going on in the organization. Things you should watch out for. Things that are going well." It used to take me two or three days to put those reports together, and I still couldn't talk to everyone in the org or follow up on every project. But I was traveling last week, and when I came back, I received this CFO scan from ClickUp Brain, our AI. It knew my role. It knew what's top of mind for me. So it gave me the bright spots I need to know about, the things moving in the right direction, and the things I should watch out for. So I'd say with something like this, yes, the work existed before, but the quality AI can provide is totally game-changing.

Chapter 9: Build vs. buy vs. wait: how to avoid AI vendor sprawl (11:44)

Russ: Actually, if you think about your own team, how have you thought about it? "We have this AI need, or something AI could possibly do. Every day there are 12 new startups. Should I look for a vertically focused AI solution just for this? Should we see if we can do it with Claude? Should we try to build it ourselves?" There's no single answer to that. How do you approach it? If your FP&A team has some new request, is it "go try to vibe code it yourself," or "let's check out this new vendor"? Do you worry about vendor creep and want one platform? How do you think about all of that?

Dan: I think the challenge now isn't which AI tool we should use. It's that we have way too many AI options. So my first question to my team is always: great use case, but does our existing platform already provide that, or do they have it on the roadmap and are about to provide it? A lot of the demos we see are just one single workflow or use case out of the hundreds and thousands of workflows a finance team needs to do day to day. One recent example: I was shown a demo of an AI accounts payable solution from an early-stage startup. Very promising. Basically no-touch; you can get it done. And then I asked the question, and the team said, "Oh, guess what? The Coupa team just showed us that they're going to launch it next month." And you already have the underlying technology infrastructure support. You have confidence in the compliance of a well-established vendor like that for finance use cases. That's a big consideration for me. So I said, just turn it on, and see if you get the same ROI.

Chapter 10: AI vendor consolidation is coming: what the Morgan Stanley CIO survey signals (13:30)

Russ: I'm curious how you think about this. You and I aren't VCs. We work at companies that have raised money, but we're not VCs. But a thing you'll hear again and again when you talk to top AI VCs is, one, you're seeing companies growing faster than any company has ever grown. And two, "Man, we're not sure if half the revenue is just going to go away, because there's so much competition from all the other points of entry." If you had to predict into the future, do you think three years from today you're working with more AI vendors than you are today, or fewer?

Dan: I think there will be a trend of consolidation on the existing platforms. Think about it: everyone has access to this AI technology. It's not like the pre-AI SaaS companies can't build AI technology on top of what they have. I'd say the platform SaaS companies are rolling out AI features at a speed I've never seen before.

Russ: Yes. Some of the developers, the good developers, are just much more productive.

Dan: Yes. But I'd say, of the three types of ROI I just described to you, the first one, automation... I think a lot of AI workflow things will get consolidated. But the second one, new capability, and the third one, the significant quality boost, might come from new vendors. Especially new capability, because it doesn't exist on the platform. You need a completely new lens of product-market fit to build the technology from the ground up. So I'm very open to seeing new players coming from those two buckets.

Russ: But at its core, you think we'll go back to what software purchasing was like three, four, five years ago, where it will just be harder for a new vendor to break into a company, as it historically has been, because the incumbents that are there are frankly building their roadmaps more quickly? I don't want to put words in your mouth.

Dan: Unless you have something new.

Russ: It'll be like the old days, where you need something really unique.

Dan: Exactly. And recently I think Morgan Stanley put out a CIO survey asking CIOs to stack-rank all the types of software they want to consolidate, and number one was application software. All these CIOs, who hold the massive budgets at real companies, are looking for consolidation. So I think that's the trend.

Russ: I recently heard an interview with Bill McDermott, the CEO of ServiceNow. He was pitching his own book, as we all do, by the way. I think Larridin is great. I'm sure you think ClickUp is great. And he was talking about how their pitch in AI is very much about the system of record, obviously, which everyone is going to say, but also really about being the best platform. His perspective was that it's just not scalable in the long run for people to go from "now I have 70 new AI vendors" to "I have 200" to "I have a thousand." It's just not going to work that way, because at the end of the day we still have data issues, we still have CIOs, we still have CFOs. Companies are still companies, and the platform is going to matter. It sounds like you would tend to think that's right.

Chapter 11: Why context sprawl kills AI outcomes before they start (16:31)

Dan: We've been deeply researching this problem for seven years, since day one of ClickUp. Before ClickUp, with all the point solutions, there were something like 17 different ways you could collaborate within a team. There's one thing that's killing business productivity, and that's work sprawl. Work sprawl leads to context sprawl, and that's why teammates struggle to find the right thing at the time they need it. And context sprawl means AI can't do the right work, because context is all AI needs to give an accurate outcome. If you don't have your context, with high quality, in one place that's easy for AI to find, your AI is going to fall flat. So that's what we see.

Russ: For a long time, the joke in media was that the only two strategies in media are bundling and unbundling, and software has been similar for the last 30 years. On the one hand, I have all these point solutions, and you go through these times where folks like yourself buy all the cool point solutions that exist, and then you get to a point where that's just too much. There's a reason Microsoft is worth $4 trillion. That's just too much; what I really want is something that works very well together.

Dan: Yes. It's ironic. A lot of those productivity solutions are actually anti-productivity. They slow people down.

Chapter 12: Change management and the challenge of moving the middle (17:57)

Russ: So I'm curious about the dynamics. I think a lot about change management. As I've said before in other forums, we're doing an unusual thing now, where we're trying to retrain the entire workforce of knowledge workers globally. We've never done that before. Everyone in the world, whether you're in business in India, in Pakistan, in Chile, wherever you are, is trying to use ChatGPT or one of its competitors, all around the world, all at the same time. And I know you said HR reports to you. It's probably a little different at a company like ClickUp, my guess would be. But as I talk to very large, very old, established companies, a thing I'll hear again and again from heads of HR is, "Look, our average employee is a 44-year-old associate brand manager. They're honestly just scared of AI. They don't want to do the wrong thing. They don't want to get fired." They're scared, A, that they'll make a mistake that costs them their job, or B, more importantly, that AI is going to take their job. I don't know if it will or it won't. I'm sure you have some employees who are 100x more productive now, and other employees who look at ChatGPT or whatever and aren't really sure what to do with it.

Dan: Yeah.

Russ: So how do you think about HR's role in change management and AI in your company?

Dan: One thing you just mentioned really resonated with me, because that's exactly what we see, and we like to use this phrase: AI cannot move the needle unless you can move the middle.

Chapter 13: The live offsite that won over skeptics, and why HR, finance and IT must work together (19:24)

Dan: I think that's the collective job of the HR team, the people team, with the executive team: to make sure we're moving that middle group to the future state.

Russ: Yep.

Dan: And we do see the same pattern. In the company, you have the AI-forward people, and you have the resisters, who are burying their heads in the sand and secretly hoping this will all blow over at some point. But you also have this massive middle. These people are capable and curious; they're just unsure where to start. So I think it's HR leadership's and executive leadership's job to lower the activation energy for this middle group. I'll give you one example. Recently, one AI-forward pioneer on my team put an amazing semantic layer on our financial data, and now he can just ask AI all kinds of analytical questions, any you can think of. He got so excited, he sent a note to the team: "Hey team, look at this cool AI answering agent that I built." And to my surprise, it was met with a lot of skepticism. "How do you know it's giving the right answer? How can it handle edge cases? I don't think AI can handle the complex business logic," and blah, blah, blah. So I sensed something wasn't right, and I discussed it with my HR partner: how can we change the dynamics here? And she said, "Time out. Let's get everyone in person. Let's do an offsite, and let's do a live test." Because these middle people, they don't respond well to hype, and they want to see concrete proof. So that's what we did. At the offsite, I asked all the human analysts to play the role of challenging the AI. They tested the edge cases. They challenged the assumptions. And gradually, I started to sense the shift in the energy in the room, from "I don't think AI can do my job" to "Holy... I can test my hypothesis in seconds instead of hours." Instead of writing queries, now I can just do the thinking. And that's how the dots were connected. Like I said, they don't respond to "look at this cool thing." They respond to "now I can see that AI can help me 10x my own productivity."

Russ: Well, also, look: confirmation bias is a real thing, and people don't like to believe they're not on the bleeding edge. We were talking about this earlier. When Netscape went public, and the first dot-com wave grew, there was a set of people who moved out here, and it was the new gold rush. They really wanted to believe they were building the future. And there were a lot of people who wanted to believe that, no, they weren't getting left behind; it was all a bubble. And of course you see the same thing now. The other weird thing I've found is that a lot of the people who talk about all the things the AIs get wrong... really, what happened is they tried ChatGPT 18 months ago, weren't sure what to do with it, read a lot about hallucinations, which really don't exist much anymore but did for a while, and went, "Ah, I knew it was hype." And then they moved on with their day. And frankly, the bigger the company, you just have a workforce of all different shapes and sizes. I liked your way of thinking about it: you have to figure out how to raise the productivity of the middle. Of course you'll always have some subset of employees who don't really care. They're just trying to hide and not work very hard, and that's true in every company. And you'll have your supercharged people. But the way you build a great company is getting those people in the middle along. So, along those lines, for a long time we had this world where HR was HR, and IT and the CIO were their own thing. Often the CIO was the person who annoyed you. When we've talked to CIOs on this podcast, frequently, for a long time they had a weird relationship with the rest of the company, not because there was anything wrong with them, but because often their job was to say no. Not unlike the CFO, by the way. In some ways, your job is to say no. That's important. You would like to not run out of money. That's the number-one job of every CEO, not to run out of money, and the CFO is the best partner there. So how do you see the evolving role between CIOs and heads of HR, as you think about change management and technology? I've personally thought, though you might disagree, that as we get to a world where some work is done by humans and some work is done by agents... we're not really there today, but we're going there; it's what people seem to believe is possible. As you think about that, how do you think HR and CIOs should work best together?

Dan: That's a great question. To scale a company and unlock the best potential of a business, you always need people, process and systems. All three pillars need to work seamlessly, in an organic, holistic way. And if you think about the three pillars, they're actually the scope of the three C-suite roles you just described. You have the chief people officer. The CFO spends a lot of time building scalable processes, because processes drive the numbers. And the CIO is on the systems side. I'd say in the old world, they were relatively siloed, and rarely went over to the other two sides. "I built a great system. HR leader, you train the people to use it." But I think the rise of AI has brought down the walls between these three pillars. For example, like I said, my HR partner is my biggest thinking partner in rolling out something like AI, because it's such heavy lifting on the change management side. So the people side is getting into the process design side: how do we design a program that can move the needle? And the systems side, the IT side, now needs to go to the people side, because we're rolling out something so new to the entire organization. There are emotions, and a lot of uncertainty. It's much different from the old days, when it was just the next iteration of what we had. This is a total revolution in how we do the work. So the CIO needs to know what's going on in the hearts and heads of the people, so the AI rollout can be successful. I think all three leaders, all three functions, need to work a lot more closely to make sure that the AI productivity lift actually lands in the business.

Russ: Sure. Look, at my last company before Larridin, we built a software product that companies rolled out to their own employees, and I think we sold something like 35 million seats of it to 600 of the Fortune 2000. It was very, very popular. And we'd have nearly identical customers: we'd sell one of them 100,000 seats, and a year later 2,000 people would use the product, and we'd sell another 100,000 seats, and a year later 80,000 people would use it. And that's still a problem if you're the vendor. So you want to partner with these folks. But yes, I think the people who believe AI is just going to do everything in an organization tomorrow, and take everyone's job, have never actually dealt with change management and getting humans to use software. And we focus a lot on: let's figure out what good looks like. Let's understand what's happening in the org, figure out what good looks like, and push it from there. How do you allow the ideas of what good looks like to percolate up? Is it you sitting with your peers, and the head of sales saying these great things were done well? Or, like you said, a business analyst on your team saying, "We built this tool," and then, "Hey, let's make sure every business analyst knows how to use this tool"? How do you make sure you understand what great in AI looks like at ClickUp?

Chapter 14: Tailoring AI impact messaging for frontline employees, leaders, and the board (27:50)

Dan: So when you think about and talk about AI impact, it sometimes requires CFOs to use different languages.

Russ: Sure.

Dan: For different audiences. For example, when you're talking to frontline employees, you need to use language that makes it very easy for them to see the potential impact. Like, "What used to take me 30 minutes now takes me 30 seconds." They immediately connect the dots: okay, that's the impact on my actual day-to-day job. Don't talk in big words that have no impact on them. So when I interact with frontline people who are testing and experimenting with a lot of AI technology, I ask them to send me: what's your problem statement, what was your life like before, and what's your result now, using AI? Very concrete, very measurable, very tangible to them. It's been very good. Every week, I ask my team to send me use cases like that. But it has to start from a problem statement. Otherwise, it's just a hammer looking for a nail. You have a problem, and a tool to help you solve it in a much better way than before. But when I talk to leaders, for example, you can't just talk about demos and shiny objects anymore. You need to attach the AI impact to their OKRs.

Russ: At the end of the day, you're either going to drive more revenue, or you're going to hire fewer people, or you're going to not spend on the software.

Dan: Yes. Your goal is not to use AI. Your goal is to grow the business by 50%. So talk to me about how you plan to get there faster using AI, and then I'm ready to give you more budget. And there's a third audience: the board, the investors. You can't walk into the boardroom and say, "This is how I use AI to do my accruals." The quick wins aren't going to cut it. What the board is looking for is sustainable, long-term improvement in your business model.

Chapter 15: Two new hiring profiles: the context optimizer and the insights interrogator (29:57)

Russ: Has it impacted the way you think about hiring? By the way, it might be that people are so much more productive that we want more people, because they can do more. Or it might be that we just don't need as many people in X. Has it had an impact yet, or are you still in the early phases?

Dan: I think it forces us to answer a very important question. It's not how many people do you need anymore. It's what is the best use of human judgment, in your business and in your function now. So for me, yes, it has changed the profiles I'm looking for. It's not that I'm not hiring people anymore, because the business needs help, but the profile has changed. I'm specifically looking for two types of people. The first I call the context optimizer. They make sure that the data structures and business processes are clean, simplified and AI-ready. They care about semantics, metrics consistency and workflow design, so AI can sit on top of it. In the old days, I'm sure you remember, we talked about "garbage in, garbage out" a lot in SaaS software. In Salesforce, if your sellers never put in the right input...

Russ: Because of Salesforce, of course. Totally.

Dan: So data quality was important in the SaaS world, because without it, your insights and your reporting would be wrong. Back then, we hired people like sales ops or admins to make sure the data quality was right in the system of record. And I think in the new AI world, we need to hire for a similar role around context quality. The job of the people I just described is to protect and optimize the context. So that's the first profile. The second profile I'm looking for is the insights interrogator. They don't just take AI output and accept it. They're critical thinkers. Like the analysts I mentioned who were challenging the AI, they'll ask about the assumption behind the assumption. They'll bring up the edge cases. And they usually realize that AI can help them test their hypotheses much faster, but they own the judgment. So I'd say that in a lot of roles, even outside finance, in any business role, we should start to look for profiles like this, instead of more tactical, functional expertise, because if AI is going to handle a lot of the mechanics of the work, labor isn't going to be the bottleneck anymore.

Russ: Yeah.

Chapter 16: Evaluating global talent on quality, not cost (32:29)

Russ: So actually, this is a non-AI question, but it sets up an AI question. If you think about workforce planning, since you're a global organization, how do you think about, "Okay, we'll hire this headcount in the Bay Area versus Turkey"? I'm picking markets where there are very smart developers in the Bay Area and very smart developers in Turkey. The actual salary costs are significantly different, but there are also coordination costs. It's easier to work with people who can come into the San Francisco office than with people who are fully remote in a time zone 12 hours away. How do you think about that at the organizational level? It doesn't have to be engineers.

Dan: Yeah. It starts with budget planning. We usually have a rough sense of how much of the budget we're going to allocate to the Bay Area, the relatively high-cost region, and how much we're going to allocate to some of the international locations. And we're not hiring all around the globe. We already have a few hubs where we know we're going to be relatively concentrated, because we have leaders there and offices there. I'd say we have two different talent acquisition funnels now. The first one is the mainstream, big talent market in the Bay Area: let's bring people in, convert them, find the right people. It's a volume game. We hire maybe 50 engineers a quarter from this funnel. And then we run a very specialized talent funnel. For example, a lot of the 10x engineers we've hired were actually acqui-hires.

Russ: Sure.

Dan: Someone who has built a product that ClickUp has on the roadmap to build next quarter. Why not hire this person, who has all the context and has already done it?

Russ: Where we pay for them, and we don't have to worry about their insurance.

Dan: Exactly. And they're self-starters. They're builders at heart. They don't need that much management anyway. They've been working from their home office or basement for 10 years. They come in, they join this large platform, and they get full autonomy and a lot of support. It's the dream job. So we see the highest retention from the 10x engineers we find through this funnel. The two funnels are complementary to each other.

Russ: But you don't have a bar where "these roles we only hire internationally, because they're half the price," versus "these roles are approved to hire in the Bay Area"? You don't really think about it that way?

Dan: We don't approach it that way. Like I said, this is not a financial decision. This is a quality decision. If it's all about ROI, and for the same budget we can find better talent in the international market, why not? And if we find great talent in the international market and we need to pay them the same as we pay in the Bay Area, why not? There's a lot of thoughtful decision-making built into hiring decisions like this. It's never just the cost.

Chapter 17: Agents as workers, and how AI rewrites corporate math on AE quotas (35:32)

Russ: Is that the right model to evaluate agent workers in the future? No one knows for sure. But in the future, you'll certainly hire some people in the Bay Area, some people in, to stick with my example, Turkey, and some agents. You might spend half a million dollars a year on a coding agent that works autonomously, potentially. So the pitches go. Is that how you imagine HR and CFOs will approach that decision in the future? For instance, you mentioned, and I won't call them out by name, because I'm not asking anything ClickUp-specific, some of the customer service AI agents earlier. In that case, you're saying, "We used to think we had to hire a person in America to answer the phone, or a person in the Philippines to answer the phone. Now we've hired this agent to answer the phone." They're not the same, but they're similar, in terms of the tickets they're closing. How do you think about that? Is it the same evaluation framework whether it's software or people, or is it different?

Dan: I think it's going to shift, and we're watching it very closely. The quality needs to be there. To your point, full autonomy and full value attribution is the holy grail, and we're looking for solutions there. And I do feel AI will rewrite a lot of well-accepted corporate math. One example is the enterprise AE. Traditionally, we know enterprise AEs carry a $1.2 to $1.5 million quota. They cost $250K, or whatever: a 5-to-1 ratio. That's the corporate math. And what's going into that corporate math? They were limited by how many meetings they could book and run.

Russ: You can only write so many proposals.

Dan: Yes, totally. It was limited by human bandwidth. And they had to do some of the BS work, like putting data into Salesforce: admin work. But now, with AI, they can have a lot more meetings booked, with AI SDRs. Their processes are much faster. They can literally spend 100% of their time selling, not doing admin work. It's already happening. So if you have enough demand for your product, they should be able to sell 40% more in a year. The productivity math should be totally reset. I'm not saying AI can replace selling. I don't think it can.

Russ: That seems like the hardest thing to replace.

Dan: Right. And it has a ripple effect. What about sales engineers? In the corporate math, you used to have a ratio, 2-to-1 or 3-to-1, depending on how complex your product is. Now with AI, I've seen a demo product that can spin up a customized demo in minutes. So that changes your baseline again. What about CSMs? What about the help desk and back office? The entire math will change.

Russ: I mean, we're seeing it today with coders, with product managers. The ratio of product managers to designers to developers is no longer what it was.

Dan: It's been rewritten as well. In my CFO circles, we talk about how some reps have already figured out they can consistently close a million, not a year, but a quarter, because they found a way to 10x and amplify their activity level. So we're monitoring very closely, and I think the winning companies will be the ones that can bend the curve on those things.

Russ: Well, this is why I've never bought that there will be less employment because of AI. Sure, there could be for companies that are just trying to cut costs, but the winning companies are going to say, "If my employees can do four times as much, I want four times more employees."

Dan: Of course. Why can't we grow our revenue 100% instead of 50%? Growing the upside is so much more fun than cutting costs, for most of us.

Russ: Well, look, especially in growth companies. It's one thing, I guess, if you're a very old software company, or a very old company inside PE that's growing at 7%. If you can figure out how to cut half your costs... They've always been run like that. Honestly, frankly, their whole goal was to move as much of their labor as possible from the US to a cheaper market in the first place. So if they can move it to agents, of course they'll try to do that. But that's not most jobs.

Dan: Yeah. And if you look at companies in the 60s or 70s, we're always surprised by how many telephone operators they had, how many manual secretaries, or elevator operators. Why did you have roles like this? I think in five years, or even sooner, when we look back at the time now, we'll be equally shocked.

Russ: Although I think it would be very interesting... I should go out and see if I can find someone like this, but it would be interesting to talk to someone who was a CFO in the 70s when Lotus 1-2-3 was invented, because it's not like Lotus 1-2-3 was invented and three days later everything was spreadsheets. But also, there's no one on your team who's allowed to say to you, "Oh, I don't use Excel."

Dan: Of course.

Russ: So that's where we'll be with these AI tools. I was talking to someone the other day, and they said, "Well, do you fire people who don't use AI tools?" And I said, "Well, you know, you could fire people who don't use AI tools, but it just won't exist." There's no such thing as a seller who doesn't use Salesforce. There's no such thing as working in accounting and not using Excel. Some people are better at it and some people are worse at it, but it's not a thing. It's just how you work. Same thing with email.

Dan: And there will be a boom in the creation of new businesses and new companies, and they'll hire the people who can [unclear].

Russ: Well, and it's hard when you're in these bridging times. At the beginning of my career, and probably the beginning of yours, it wasn't that unusual to hear about CEOs of large companies whose assistants printed out their email for them, and they'd mark it up and the assistant would type the reply. But that was a very short period of time, when you had some much older, very senior folks and the tech was new to them. It wasn't the norm, and within five years it was gone. If you heard that today, it would be among the craziest things.

Dan: You were the bridging generation.

Russ: Yeah. Well, there's always a bridge to the next thing. There's always some generation bridging to the next thing. Frankly, that's why HR and change management matter so much.

Chapter 18: Why curiosity and humility are now the defining hiring criteria (41:52)

Russ: So actually, I'll end with this last question. As you think about this, how do you think about what you're hiring for with young people? I know you have children; we were talking about it before the podcast. I have children, and I won't ask what advice you'd give children or anything like that.

Dan: I've probably become very irrelevant to my children.

Russ: Yeah, of course. The technology is changing so quickly. But I am curious: when you're hiring younger folks at ClickUp, do you find what you're looking for today is different than what you were looking for five or six years ago, or different from what your first boss was looking for when he or she hired you at the beginning of your career?

Dan: I think the fundamental traits are the same, and those are the traits that will determine if this person can shine in the age of AI: curiosity and humility. I'm still looking for those two things. And for the two roles I described before, the context optimizer and the insights interrogator, the underlying trait is: are they curious? If AI gives you an answer, do you just take it and run with it, or do you say, "Hey, what about this angle? What about that angle?" Do you challenge the AI's thinking? I've seen people do that, but they're not the default. A lot of people see a first round of AI output that doesn't look right, and they move on: "AI is not good." But these curious people turn AI into a thinking partner. They test their own thinking. So now, when I give cases to the people I interview, the first sentence is, "Yes, you can use AI, but show me your prompt, and let's do a live session together. Let me see how you iterate and keep trying different angles to get to the best answer you can." Previously, it was, "Don't use AI; that's cheating." I just don't think we can be that narrow anymore. Embrace it, but it still tests the underlying intelligence and curiosity. And are you humble? I think a lot of arrogance comes from "I don't think AI can do my job. I think I know more than AI." Your context, your domain expertise, your knowledge are all going to be commoditized. Can you be humble, and see that you can learn new things? I think that's a determining factor in whether these people can be successful in the new world now.

Chapter 19: What young people should study in the age of AI (44:02)

Russ: And I guess, last chance, actually: do you want to try giving advice to young people on what they should study if they're going to college? This should become a recurring segment. I should ask people this all the time. Everyone talks about it.

Dan: I think communication, original writing and critical thinking become even more important. Why communication? Because AI can write all this stuff for you now. I observe how my kid uses AI, and I'd say that whenever he struggles to describe what the end state looks like, or what exactly he's looking for, the AI is bad. But whenever he clearly has a vision, the AI is 10x; it gives the kid wings. So I think we all need to train our ability to communicate precisely and sharply, and our ability to articulate a thing, an emotion, a thought. It's going to be your only way to bring AI to your use in the future.

Russ: This has been fantastic. Thank you so much for joining me.

Dan: Thank you. It was so much fun. Thank you, Russ.

This transcript was prepared by AI and may contain mistakes.