On this episode of the At the Edge podcast, OpenAI CFO Sarah Friar joins McKinsey senior partner Lareina Yee to talk about how to lead through this era of intense technological adaptation and what it means for the future of the finance function and business leadership.
This transcript has been edited for clarity and length. For more conversations on cutting-edge technology, follow the series on your preferred podcast platform.
What the finance team of the future looks like
Lareina Yee: You have held leadership roles at companies from Salesforce to Square to Nextdoor, and now you are leading finance at OpenAI, where using generative AI (gen AI) to automate activities like many of those within the finance function is a core conversation. What has that experience been like? How are you envisioning the finance team of the future?
Sarah Friar: It’s really fun to get to work on building a platform of the future, where you have a tool that can do increasingly complex tasks and empower people to work more efficiently and creatively.
Subscribe to the At the Edge podcast
What that means in real life now in our finance function is that we’re using ChatGPT to do things like unify data from different sources and code AP [accounts payable] invoices. That shortens our close, which is something all controllers think about.
Investor relations is another good example of where we’re using gen AI in the finance function. We just finished a financing round, and in the middle of a deluge of in-bound diligence questions, we were feeling underwater, so we built an investor relations custom GPT. We fed it the knowledge of all the diligence questions we had answered up to that point, and we fed it our management presentation. We also told it not to look externally for answers, as there is a lot of incorrect information published about OpenAI. And now we have an investor relations GPT that allows us to answer questions in seconds that previously took hours or a whole day.
Things like that have changed my life. Also, the technology has expanded the skill set of the team. For example, in finance, it’s very useful to have someone who can write code or help with SQL [structured query language] queries, but that is not a common skill set in finance. Instead of asking for help from our technical organization, we can now just ask ChatGPT to assist in writing that SQL query. This has really advanced our team from number crunching to being a better business partner.
Today, there’s still so much in the finance department that is the look back, not the look forward. I hope in five years we’ll look back at how we do things today and feel like today’s methods are so outdated.
Lareina Yee: If all of that is possible now, what will the finance function look like in five years? What things do you hope you’ll be able to do with gen AI?
Sarah Friar: I really hope we will move completely to the point of being the home for business insights, driving the business to go bigger and faster. Today, there’s still so much in the finance department that is the look back, not the look forward. I hope in five years we’ll look back at how we do things today and feel like today’s methods are so outdated. I want to be able to look at my team and see that everyone is in that mode of forward-thinking and insight-driven work, and that a lot of the work we do today becomes rote work that gen AI does for us.
Be in a position of impact
Lareina Yee: You’ve had many different roles in your career. You were recently the CEO of Nextdoor, and before that you went from being the CFO to the COO and CEO at other companies, and now you’re back to the CFO role. Can you talk about some of these decisions?
Sarah Friar: What I often tell people I’m mentoring is to forget about titles—just do your best work and get yourself into the places in a company where your skill set and the ability to have impact that matters overlap. You want to get to where you can have the most potent impact.
The reason I made the shift to OpenAI is easy: I’m in the crucible of the AI transformation. It was not a hard decision. Even my kids were kind of impressed and interested, because everybody’s talking about what’s happening at OpenAI. To get under the covers and actually see where the research is taking us is such a privilege. Another thing that is important to me is being a leader in a human sense—it’s important that we have diversity around these tables as we build technologies that are literally world changing.
As for the CFO role, it’s still probably one of my favorite jobs because “CFO-ing” is about driving strategy and empowering people. When I’m doing my one-on-ones here at OpenAI, and I’ve done more than 80 already, I always ask, “What can I do to accelerate you?” That is the most important thing for me to take away from every one-on-one I do with my colleagues. I like being the person who is figuring out how to say “yes.”
How do we build a community where it’s a world of possibilities and opportunities rather than something that can be more fear-driven?
Lareina Yee: So the title is not actually the objective. But there are certain sets of capabilities that you’ve acquired over time that allow you to be at the center of driving substantial change. Can you tell us a little bit about some of those capabilities that are timeless, irrespective of the shifts in technology, including the advent of AI?
Sarah Friar: First and foremost is knowing how to break down and solve problems, and how to communicate. Next is being strategic. A lot of people don’t really define strategy well. Their tactics become their strategy, as opposed to having a strategy that they put alongside tactics. I think these two skills have always helped me in whatever role I’m in to really help drive a business.
And then I’d say the skill of understanding where community meets kindness and humanity is important, so I’m not losing who I am to the circumstances of my work. I actually asked ChatGPT about this the other night, when someone I was with had me type in the prompt, “Based on what you know about me, ChatGPT, what’s the thing I might not know most about myself?” What an interesting prompt, right?
And the answer it came back with was about how much growing up in Northern Ireland still continues to shape the person I am today. I love that answer, because it reminded me that the culture of where I grew up really is important. For example, I see how my parents’ investment in their community comes back full circle now that they are the older generation and people in their community check on them. They feel really supported by it.
So I think about the community I’m building in my work in relation to that. Maybe it’s my community in my company or the community of my ecosystem. Or it’s how we at OpenAI show up in the world where we become synonymous with AI. How do we build a community where it’s a world of possibilities and opportunities, rather than something that can be more fear-driven?
AI as enabler of both business and democracy
Lareina Yee: You recently published an essay in the Stanford Digital Economy Lab’s Digitalist Papers, where you write about how technology can amplify and increase kindness and community connection. That might be a counterintuitive thought for many people. Can you tell us about that?
Sarah Friar: My coauthor Laura [Bisesto] and I felt strongly that if you think about what creates a healthy democracy at a grassroots level, community connectedness is the bedrock. When people are connected, they have trust, and then they are able to embrace the idea that they live in a democratic society. There’s a fairness to how things are done. So when you’re trying to build community and trust, communication is everything. AI can really help with reducing miscommunication. When I was CEO of Nextdoor, we used OpenAI’s API to help rephrase remarks on our platform that could be perceived as hurtful. We found that popping up a kindness reminder and rewriting a potentially offending message with gen AI could be powerful.
When communities are healthy and wealthy, things like democracy tend to flourish more.
Also, there’s a business element to every community. What makes a strong neighborhood? It’s not just neighbors. It’s the whole ecosystem. It’s the schools, the churches, the sports teams, and definitely the businesses. I’ve got a total soft spot for small businesses, particularly those started and owned by women and nonbinary people, where the founder is everything to the business—CEO, general counsel, CMO, CFO. There is so much to be done, and marketing tends to be one of the places that really can make or break that business. We felt AI could bolster a business by helping with basic things like a marketing plan and so on. In doing that, you bring economic stimulus to a community. When communities are healthy and wealthy, things like democracy tend to flourish more.
Probably the most pertinent piece is government. How can government use AI for better personalization and one-on-one communication with its constituents? We talk today about voting blocs, as if this homogeneous big group of society all does the same thing. They don’t; we have different needs and wants. But usually, it’s cost prohibitive for a government to treat us as individuals. With technology that uses large language models and things like ChatGPT, suddenly you can have incredible personalization.
For example, the state of Minnesota uses ChatGPT today to create increased accessibility to the government for people who may not speak English. In automating all that translation, they’re saving hours of people’s time and hundreds of thousands of dollars in costs monthly. And they’re creating a one-to-one experience, where if I am a refugee or a recent immigrant who needs help to get on my feet, which often includes building a business, the state is now able to do that in a much more personalized way. So those are tactical examples of how we feel AI can improve the bedrock of democracy.
How companies are using gen AI now
Lareina Yee: Let’s dive into the innovation going on at OpenAI. How do you see gen AI reshaping business in the future, and what use cases are you seeing now and in the future?
Sarah Friar: I was just in New York with customers, so I’ll share some thoughts from that. Morgan Stanley is a great example of a platform that’s organizing their huge knowledge base, particularly in areas like research, and using it to automate operations. They use it to do a better job of getting information out to clients, by generating summaries of video meetings and drafting follow-up emails. With that kind of automation, wealth managers can focus more on customers, rather than having to think first about the steps they need to take to send customers follow-up emails.
Klarna is another. They’ve been very loud and proud about how their new digital-shopping system built on our API is helping customers find the right products at the best prices, and also how much they’re saving on customer service. And Mercado Libre was at our event last week, so I got to hear their CTO say to the whole crowd how they’re using ChatGPT to autonomously manage customer service decisions. That involves about $450 million annually on our platform, so that’s a lot of money that is being touched by our technology, and also cost savings.
Square is yet another. Their customer service team has one of my favorite objectives: “turn questions into commerce.” So you have an inbound question from a customer who needs something fixed, but you also use that interaction to provide them with added value that helps their business thrive and bring in more revenue. The more we think about customer service as moments where you’re going to help the customer do even more, it’s a win–win. For businesses, AI is already helping do this.
Lareina Yee: OpenAI launched new reasoning capabilities recently, and that brings us to the topic of agents and what is called “agentic technology,” where the AI is making decisions and learning from experience to achieve goals without direct human supervision. Can you help us understand how the new reasoning capabilities lead to agentic technology?
Sarah Friar: We launched a new set of models with reasoning capabilities: o1-preview and o1-mini. This is how we characterize this path ultimately toward AGI, or artificial general intelligence, which is our mission: AGI that benefits humanity. The reasoning capabilities eventually become agents, or agentic technology. With this, the model can reason in places where historically we would have tried to code in software, but it involved too many exceptions and so reached dead ends.
The really exciting next thing … will be agentic innovation, where you’re contributing to new knowledge in the world.
With agentic technology, the AI can take actions in the world and make some decisions for you. It might be something as simple as, “Order something for dinner tonight that my family might eat.” And then you can extrapolate that kind of decision making out to longer-duration things like a credit process at a bank—it goes from a customer having a need to money in their bank account that they’re now spending to grow their business.
The really exciting next thing after that will be agentic innovation, where you’re contributing to new knowledge in the world. When you hear Sam [Altman] and other folks at OpenAI talk about doing things like curing diseases that we have not been able to tackle, or helping solve climate change problems, this is the moment where innovation is happening. And we’re on that continuum.
Improving ways of working with gen AI
Lareina Yee: What does that mean for how teams work together in our workspaces?
Sarah Friar: It is going to be fascinating how the world of work changes, especially when you add in multimodal communications. With advanced speech today, for example, we’ve rewired ourselves to talk with our thumbs; certainly my teenagers are experts in talking with their thumbs. But now we’re moving back into a world with multimodal communication, where interacting with technology can feel more human because we’re using different methods together to communicate.
This can be phenomenal, for example, for older generations. My mom has really bad macular degeneration, so she cannot type with her thumbs, nor can she read most things coming in on a small-screen phone. But if she could interact with technology verbally, that’s just a more natural way for her to communicate given her limitations.
Lareina Yee: Like during the pandemic, we changed the physical location of work, but we didn’t totally change modalities and times. Moving to multimodal communication could change a lot of assumptions that we make. And you could have AI agents as part of your team soon. Our relationship to machines, just like our relationship to our phones, could be very different.
Sarah Friar: If something like COVID happened again and you had the AI we are about to have, as a CFO, instead of calling your team and saying, “We need to scenario plan really fast,” you could pair reasoning to your financial model and talk to it in a natural way and say, “Hey, GPT model, what would happen if no one walked into my store for six months? How would my company survive?” On my own as a CFO, I can think about what scenarios might play out if my revenue went down 10 or 20 or even 40 percent. But being able to interrogate a model with a natural language overlay would be incredibly powerful for CFOs everywhere.
Lareina Yee: In a crisis, to your point, what you want is your people taking action that only humans can do, and not doing the scenario planning that a machine could do.
Sarah Friar: Exactly.
Being the rising tide that floats all boats
Lareina Yee: Let’s shift gears and talk about Ladies Who Launch, which is a nonprofit you cofounded to help empower women and nonbinary small business owners by providing really practical things like grants, mentorship, and networking opportunities. What inspired you, and what have you learned from this experience?
Sarah Friar: The inspiration came from an event we had at Square—I had just joined the company, and we wanted to do something special for International Women’s Day. We decided to invite some of the sellers [who use Square] into our four walls and have an evening with them. I was not worried at all about whether people would come to this event because [Square cofounder] Jack Dorsey was our keynote. I thought, “Everyone wants to talk to Jack. This is perfect.”
But what I realized that evening was that, while Jack was awesome, what the women and nonbinary individuals who were there really benefited from was, number one, just finding each other. When you’re in a minority, you recognize how hard it is to walk into a room and see no one like you. When you’re in the majority, you just don’t even notice. So, first of all, we created that density so they could find each other.
The second thing we realized was the importance of community building and education. Yes, it’s great to hear from someone who has built massive businesses, but the sellers wanted practical tips from people who are in their shoes doing the same thing. They really wanted to hear the small business owners up on stage talking about how they had dealt with creating a social media marketing campaign or building a business plan or getting that first financing.
The third thing we learned was that the inspiration the sellers got from hearing each other was so powerful. When women talk, they tend to talk about their failure moments. It’s a real humility and authenticity that’s visceral. For example, we had another event where [fragrance entrepreneur] Jo Malone came to speak. If you’ve ever smelled a Jo Malone candle or walked through her displays at an airport, you know how amazing her scents are. But you may not know that she had cancer and lost her sense of smell for more than a year to her chemotherapy. So for someone whose entire livelihood and sense of self was literally their sense of smell, that was incredibly challenging. For her to talk about it publicly was one of those powerful moments. I remember women afterward saying, “If she can do that, I can do anything.”
Go for it. You might fail, but you’ll learn from it. Absolutely go for it.
Lareina Yee: Based on those experiences and your own career, what advice would you give a young woman who’s a budding engineer or who has maybe been in the workforce for a while and is thinking of trying something new?
Sarah Friar: Number one, go for it. You might fail, but you’ll learn from it. Absolutely go for it. Second, network like crazy. It’s a great way to learn and make something happen. Every time you meet with someone, always end by asking, “Who are two other people you would introduce me to?” It’s forced multiplication of your network. There may not be something in that moment, but things come together with time. I’ve found there’s a serendipity, where a year or so later, something comes up and I remember, “Oh, that person who I met.”
Also, women often suffer from believing they have to do something for you first to then ask for something from you. But the research shows that if I ask for something from you first, you’re actually way more invested in my success. So make the ask. It’s how you’re going to get things done.
Lareina Yee: That’s actually good advice for both men and women. Let’s turn now to you personally with some short questions for you. I already know that you are great at prioritizing exercise and sleep. I don’t know how you do that, but you do. Are you a night owl or a morning person?
Sarah Friar: I’m a morning person. It is when I have the most clarity and energy. I love nothing better than to be out on a walk when the dawn is coming up. It’s when I have all my best thoughts.
Lareina Yee: You were a rower at university. What do you miss about that?
Sarah Friar: I miss being that fit. And I miss the discipline and teamwork that comes with being on a rowing team. In a boat, you can’t compete with the other rowers in that boat; you’re there to complete the whole [set of] eight [rowers] and, in particular, your pair within that eight. Rowers will know what I’m talking about: if your pair is slightly tipped down, you need to tip the other way to balance the boat. If the oars are catching a little slowly, you almost need to bring them into sync. I was always the seven seat, which is a really interesting place in a boat, because the eight’s the stroke and sets the pace, but the seven takes the stroke’s beat and then translates it for the rest of the boat.
There’s a huge confidence in the seven seat that—no matter what this person in front is about to do—the seven seat is mentally sending the message down the boat, “Don’t worry, we got this. We are going to be able to keep up.” And I still love that feeling to this day. I get it a lot at work, but I really miss just knowing I’m going to get in a boat and that feeling is just going to manifest.
Lareina Yee: And my guess is that boat was being rowed early in the morning, so it all fits together.
Sarah Friar: It all fits.
Lareina Yee: Seeing you as the person who completes the boat—I think that is really such an encapsulation of your leadership. Sarah, thank you so much for joining us.
Sarah Friar: Thank you, Lareina. It was such a pleasure.