What's for Dinner? AI-Driven Commerce with Instacart CEO Fidji Simo

0:05Welcome to KnowPriors. Today we're speaking with Fiji Simo, the CEO of Instacart, the leading grocery technology company and co-founder of the Metrodora Institute, whose mission is to cure neuroimmune access disorders. She's also the former leader of the big blue app, Facebook. Instacart is transforming the way we shop, making it possible for millions of busy people to get the groceries they need from the retailers they love. We're really excited to explore Fiji's refounding of Instacart, their use of AI, and the impact of AI on commerce and healthcare. Fiji, welcome to No Priors. Thanks for having me, Sarah and Yelena.

0:36Let's start with some personal background. You've been a star leader in technology for a long time, from strategy at eBay to meteoric rise to product and eventually beloved leader of Facebook and now CEO of Instacart. That is a wide range of different things. What's driven your career decisions? Wow, that's a hard question to start with. I think it sounds cliche, but it was all about where could I have impact on problems I deeply cared about? and I always say that I am a pragmatic technologist and what I mean by that is that I love technologies that solve real problems for real people and eBay was a good example of that you know when I joined the company I was kind of in love with this idea of like all of these like communities of collectors exchanging goods but really exchanging their passion and I was fascinated by that.

1:30And then, you know, my transition to Facebook was very similar. I was an immigrant in the US and I was starting to see that it was so much easier to stay in touch with my family back in France, thanks to Facebook. And then transitioning to Instacart was kind of similarly about the mission. You know, I thought the idea of connecting people to the retailers they love so that they can enjoy the food they love and spend more time with their family was so critical and food is, you know, one of the most important things in our lives. And so it felt like I could have a big impact there as well. Yeah.

2:07Amazing. I remember what an impression you made on me when we first met, when you were at Facebook, like you are very French, you wear lipstick and heels. You've led some of the most iconic products and companies in the tech world, but you don't subscribe to some of the, let's say Silicon Valley monoculture. And you were clear, very conscious of that, like, why is that important to you? Or how do you think about like cultural image and how you present as a leader? Well, you know, I've been French since I was born. I have been wearing lipstick and high heels since I was 12. So it is such a big part of me that it was never a conscious decision.

2:45It was just about being me. And then when I arrived in Silicon Valley, a lot of people were commenting on that. And I was like pretty surprised because, you know, that was just me being me. And they were telling me, well, you know, especially as you move into product development, you really need to blend in with the engineers. So like remove the makeup, tie up your hair, start wearing a hoodie. And I did that for one day at Facebook. No one recognized me. I would pass in the hallways. No one would say hi. And the reality is that it felt like a lot of effort and work not to be you. And when I thought about it, I realized like, you know, if I have a lot of effort and energy to put into something, I'd much rather put it into building amazing products than trying to shape myself into something I'm not.

3:31And so, you know, I decided to embrace the French accent, embrace the heels and all of the things. And, you know, people around like adapted to it and then got used to it. And that was not a problem anymore. So, you know, So better to be you, basically. I asked a mutual friend of ours, Ravi Gupta, what I should ask you. And he said that you are, you know, a very, very good leader with tremendous loyalty from your people while still being really tough. His suggested question was, how do you, how do you earn followership as a leader? Oh, that's a good question. Not surprised Ravi would ask that.

4:11I think, you know, I tell my leaders when I start, you know, working with them and I did that when I joined Instacart that I want to be like, I want them to look back and realize that I was a big accelerator in their career. And I'm telling them that to play that role, like it's a partnership, obviously they have to want that. And for that to happen, it's going to mean two things. it's going to mean that I am going to hold them to a very high standard and give them a lot of tough feedback and help them meet that standard. But it's also going to mean that I'm going to be there backing them up every step of the way and being in their corner and wanting them to succeed like no one else has wanted them to succeed before.

4:59And so I think, you know, it's not for everyone. Like there are some people who are like, whoa, that sounds way too stressful. Thank you very much lady i'm gonna opt out but that's actually a good you know a good thing because like i don't think you need to be necessarily a leader for absolutely everyone you like you pick your team and they pick you and the people who like that end up being people who want to excel uh but want to excel through uh a leader that like is gonna back them up is gonna be compassionate is going to create enormous safety for them to stretch themselves and potentially fail.

5:35And when that happens, you know, across many people in the team, you end up creating exceptional results because everybody kind of is in this mindset of like doing the best work of their lives. I imagine becoming CEO of Instacart at the time you did having to like, you know, learn to be CEO, build that followership, recruit the people you need, but also at a time of like massive, somewhat unexpected growth for Instacart and becoming critical infrastructure with the COVID-19 pandemic was a lot. How did you react to all of that? How did the pandemic change the company? Yeah. I mean, what do you mean, Sarah?

6:17It was a total walk in the park. It must have been chill. Very, very chill. Well, listen, first off, 10 years at Facebook prepares you for a lot. So I would credit that experience with me being ready to handle a lot of stuff. I would say I joined Instacart in 2021 and it was a very interesting time because the company had forexed the year before. We were in the midst of just absorbing all of the demand from customers, completely new habits, much larger team all of a sudden that we had to hire. but at the same time there was also this question mark on is instacart going to be just a pandemic darling a you know one trick pony and are people just going to go back to their like usual way of doing things and the company is just not going to maintain that and you know fast forward a couple years clearly we have continued to grow during this time and clearly these habits are here to stay.

7:17But what you do during times like that is like, you need to go back and focus on mission and vision. And so I spent a lot of my time really painting the picture for the company of what that company could be. And we even did like a full day called the Vision Summit where everybody needed to be on the same page on the fact that Instacart had become critical infrastructure that came with a lot of responsibility. Instacart was also a retail enablement company, meaning that we weren't just building the Instacart app. We were actually building technologies for our grocers. You know, when you go to Publix.com, Kroger.com, Schnucks.com, like that's Instacart technology that's powering a lot of that.

8:01And that it represented a huge, you know, endeavor that would give us, you know, a lot to do over the next 10 years, but a very big opportunity. And so you have to get people to rally again after like a massive period of growth like that to figure out what is the next mountain, what is the next thing that we're going to all go accomplish together. I guess related to that, it feels like one of the next mountains that you all are conquering has to do with the incorporation of AI into your product. And obviously you care more about how the product impacts the users, but I'm a little bit curious, you know, when did Instacart start focusing on AI and ML and how do you feel that this wave of generative by differs.

8:39Yeah. You know, it's interesting because obviously AI is having a big moment, but Instacart has been focusing on AI since the very beginning. Like if you think about it, we are fundamentally a search engine. We do a billion searches a month, a quarter, sorry. And you know, it's all powered by AI. Every time a shopper goes to the store and a product is not on the shelves, we suggest replacement. We have suggested a billion replacement at this point. And that's all powered by AI. The way in which we match shoppers to go to specific stores, all powered by AI. So the list goes on. But I think the thing that's changing now with generative AI is that we are finally at a point where we can answer people's questions using natural language and the way they've been asking themselves these questions before.

9:30If you take a step back and you think about commerce online, it is very bizarre that we have forced humans to train themselves to express their commerce needs in the form of keywords. I mean, people don't think about how to put dinner on the table for the week in terms of cilantro, eggplants, sour cream, which is exactly what they're currently typing on Instacart right now. Instead, what they think about is like, I have a budget of 300 bucks. I have to put, you know, X meals on the table. My daughter is lactose intolerant. We traditionally eat Mediterranean food and I have a set of recipes that I always cook, but I'm open to suggesting, like to trying out a new one.

10:20Like, let's figure it out, you know? And that's the way in which, you know, like people ask themselves this question and therefore that's the way software should answer them. And to me, the thing that's most exciting about this new wave of generative AI is that we can finally match the answers that we give people to the way they ask themselves these questions. Yeah, it makes a lot of sense. You know, there has been this sort of shift in terms of what the underlying technology can do. And I feel like so many people just kind of extrapolate, hey, this is just the same AI curve as what we did with CNNs and RNNs and all those sort of prior waves.

10:54But in reality, it's a new technology, just continuity with a new underlying model and new capabilities. How did you get the team to start thinking in terms of those capabilities? Because I feel that often when enterprises now adopt generative AI, some of them don't actually realize that it can do dramatically new things. Yeah. So a couple things. First off, it goes back to my philosophy as pragmatic technologist. I don't get enamored with the tech. I get enamored with what he can do. And so I actually, the prompt I gave the team was, what are the problems we've always wanted to solve for our users that we can now solve so much better?

11:30And I see so many companies like coming at it from the point of view of like, do we need an AI roadmap? And I'm like, no, no, no, you need to figure out like, what is the biggest problem you're trying to solve? And now can this new tool help you solve this problem in a completely different way? And so that was very much the prompt. More tactically, though, we have had two approaches. One approach is a centralized approach. So under our chief architect, we now have a centralized team that is really going deep in understanding these technologies and trying to figure out what can be done that couldn't be done before.

12:07And then we have a federated approach, which is really about like taking some of these learnings and then making sure every team uses AI. And I think that has helped us well to have this combo because you can't expect every engineer to go super deep to understand everything that's going on. But at the same time, if you keep generative AI just in a centralized team, it's not going to impact all of your products. So you really need the combo of the two to make that happen. And do you have then, I guess, from a tactical perspective, I've seen some companies hackathons where they just say, okay, this week, everybody's going to experience this technology just so that people get that intuition.

12:42Because I think the intuition is missing. I've seen some companies have the sort of consultants who will drop into different product teams and try and discuss what could they actually do with the technology or what does it enable and give some almost very fast demos on it. Is there anything sort of tactically that you all did along those lines? Yeah, we certainly did a big hackathon focused on that. And again, the prompt wasn't just use generative AI. It was like, take all of the problems we're already trying to solve and figure out if generative AI can solve them way better. And that generated a lot of great ideas.

13:14And then we have this model because we have this kind of centralized team that we call Carrot AI. These people get pinged all the time by the rest of the company to kind of advise on how to deploy some of these things. We've also deployed an internal assistant called AVA so that all of our engineers can leverage the GPT-4 models in a lot of what they're doing, but do that in an assisted way that helps them figure out how to integrate that. So that has also been very helpful. And the thing also is like, it reminds me a lot of like the transition to mobile. When I was at Facebook, I was put in charge of monetizing ads on mobile right at the time where there were all of the headlines post IPO saying like Facebook will never monetize mobile, mobile is too small of a screen to have ads.

14:05Like, you know, you remember that. And the reality is that like, you know, there were a lot of companies that were trying to take the desktop ads and just like shrink them to a mobile screen. And instead, you know, what we came up with was ads in Newsfeed, which was a completely different paradigm. And I use that a lot internally to help people understand that being AI native is going to require like really rethinking from the ground up how you use these technologies, not like trying to fit them within in your existing paradigm. And I'll give you a very concrete example. We recently launched Ask Instacart, which is a way to like, you know, ask questions like, you know, what are the best sources to grill my chicken and things like that.

14:49And there was a lot of debate on like, should it go into the search box or not? Because the search box is sacred, you know, on a commerce site. Like you type keywords, like people are not going to know what to do if we start putting questions. Let's put it as a separate thing. And then it's like, no, actually, if you were rebuilding commerce from scratch, that search box would be the place where you start asking questions. It shouldn't, like we wouldn't like think of it as a keyword search box. And so these are the micro decisions that are actually big ones to train the company that we need to disrupt ourselves and the prior paradigms in order to be truly AI native.

15:26Can you describe why you guys did an internal assistant versus just saying everybody, you know, go hit chat GPT on your own or whatever, and sort of what's been interesting from a productivity perspective so far? So we just released that. So hard to know on productivity, but it was a lot about like, you know, making it more accessible and making it easier for engineers to kind of deploy these things as a bunch of like keyword shortcuts and just making like the interface very user-friendly so that engineers would feel like it's just a core part of their work now. The thing I would say about productivity is that it's early to be able to measure any of that, but I think it's going to impact productivity well beyond engineering.

16:13A lot of people are talking about making software development easier, but I'll give you some other examples. like our marketing team, for example, created an entire library of animated characters using AI and using, you know, limb mirroring technology and things like that. And now we have like 200 plus, you know, characters and animation that we can use in marketing. That's huge. That would have, you know, cost a ton of money if that was to be done, you know, not using AI through an agency. Our legal team has used AI to figure out how to detect the specific terms, the custom terms that we have with specific retailers so that we can track that in a much better way.

16:56Like all of these things, they seem much more minor than, you know, launching a fancy new AI thing. But they actually make the lives of each team way better. And I think it will create a lot more satisfaction with the job. Because when you think about it, like no lawyer will tell you, yeah, going through a thousand documents to look for custom terms was a highlight of my day, you know. And the fact that AI can actually do that for them, summarize that for them, that means that we're going to get the best out of our people and we're going to remove a lot of the more minute tasks. Are there any other sort of big areas that you've seen either impact productivity or the customer experience or things that are coming that you can talk about that you're just excited to be launching?

17:41I know that you did things, for example, the ChatGPT plugin and you were very early there. I'm just curious if there's other parts that are sort of exciting and on your roadmap. Well, there's so much. I think the ChatGPT plugin was a great example because it allowed us to start to see how people were using these things. And again, the questions are fantastic because right now, Instacart is really a tool as of today that is very bottom of the funnel. You have to have already decided that you want grocery delivery, that you know what to eat, all of these things. Whereas when you look at the ChildGPT plugin, it starts with, I have a birthday party.

18:22I don't know which snacks to get. There's a lot of intent and different occasions that we can tap into and serve better because we can capture intent in a very different way. And so I'm very excited about that. and I think we're reproducing a lot of these experiences with Microsoft Bing, with Google Bold, and hope to be kind of the layer of conversion between bits and atoms. Because the way I kind of see this trend of generative AI is that everything software is going to become a lot easier and lower barrier to entry. But the translation between software and atoms is still very hard. And the fact that Instacart has aggregated, 80 ,000 stores, more than 1 ,000 retailers, means that we can create these magical experiences where you ask the software things in a natural language and food shows up at your door in two hours.

19:19That's pretty incredible. And so we're really doubling down on that. And then you ask what's coming in the future. In the future, I want to make that available as a platform because you could imagine any nutrition app any apps that's doing like anything that has to do with food or even commerce tapping into this network and making it available to like basically turn intent into things being like showing up at your door in two hours so we're very excited about that and then longer term i think it's all about like how do we answer all of the questions that people ask themselves and And the thing I would highlight is that it requires these large language models, but it also requires having really unique data.

20:05And the fact that we have an incredible catalog, the fact that we have billions of data points on what people need, that's what gives us our strengths. And we're marrying the strengths of these incredible data with these large language models. And that's what's turning into really unique experiences. And so my advice for any company is just really thinking about your data infrastructure and the unique data sets that you bring to bear in this new world that can create these delightful experiences. Yeah, that makes sense. I guess related to that, given that you're a real world company, a place where AI perhaps has shown a little bit less progress than the recent burst of LLMs is in things like autonomous vehicles and self-driving in terms of drones, delivery, automated picking.

20:50how do you think about the more sort of real world robotics side of this? And when do you think that's coming? Yeah, we've been following this incredibly closely, as you can imagine. And, you know, my take is if we were starting from scratch, like rebuilding the grocery industry from scratch, we would probably not rebuild it the way it is. There will probably be fewer stores the way they are. There will probably be more dark automated stores. But the reality is that we're not starting from scratch. And when you think of Instacart, like we have a network of 80 ,000 stores, these retailers are already paying a fixed cost on all of these stores.

21:30So it is actually incredibly efficient to be sending people to these stores. Like these retailers get incremental demand on top of that fixed cost base. And these stores are very close to the customer. And so a big part of why these kind of automation initiatives have failed is not because the technology is not good and not there. It's because the economics are actually very hard to make work. Like if you want to have like a very large automated warehouse, you kind of need to put it further away from the customer so that you can aggregate all of these orders. Now you're talking about next day delivery instead of same day delivery.

22:08And we all know that customers want speed. And so you end up in a situation where, yes, like you have great automation, but that's not reducing costs and that's not improving the customer experience. So back to my point on pragmatic technologies, there are some technologies that are great, but if they don't help with either cost or customer experience, you know, they are not going to get mass adoption anytime soon. Can you describe what you guys are doing with the connected store effort and caper carts? Because that's a different direction of like, you know, more extension into the real world.

22:39Absolutely. So the thing we realized is that the grocery industry is in the middle of this massive digital transformation. And that digital transformation is not just happening online. It's also happening inside the stores. And grocers really turn to us as the partners that can deliver them the best technology solution during that transition. And so when I joined, I think within a month of me joining, I made the acquisition of Kaper Carts because they are basically smart carts where you can put products inside the cart and using computer vision and AI, of course, the product is automatically detected and then you can check out from the cart directly.

23:19And so these kinds of technology, whether it's caper cards, whether it's Canon Pay from a mobile phone, these kinds of technologies are really helping grocers embrace that digital transformation, are helping connect and bridge online and offline. Because we think that we actually have data that shows that customers who are omnichannel customers are more valuable than the online-only customer or the offline-only customer. and so that that enriches the store experience and also makes customers more likely to use that retailer across those channels so we're really excited about that we have a keeper cart rolled out at Kroger, Sobeys, Wakefern and it's a pretty magical experience and it's not just the part of skipping checkout it's the part where you have a screen in front of you and now if you drop the particular item into your cart we can like suggest a related dressing and we can tell you which aisle to find that dressing in.

24:18And we can show you which savings you're getting and you can manage your budget on that screen. It's a really rich experience that stores are missing today. And I think that's going to be, you know, over the next five years, the future of grocery shopping. I have to ask because you have multiple different experiences that are really deep and interesting in advertising from Facebook to what you're doing with retailers and CPG brands today. If you zoom out, what predictions do you have for the impact of AI on advertising? Oh, I think it's going to be fantastic for ads. I think it's really about much more personalization and really like this kind of ideal that, you know, you're going to get the perfect ad at the perfect moment for you.

25:02I think it's also going to be an explosion in creative when you think about all of the variation of creative that can be creative now so that, again, we match the right person to the right user. That's going to be just incredible. And that's going to, again, make advertising a lot more relevant. And then the other thing is that I think advertising fundamentally follows the consumer products and the consumer use cases. I mean, that was one of our guiding principles at Facebook. and because we're going to see such an explosion of new consumer experiences, I think as a result, we're going to see new placements for ads that are going to be interesting.

25:40So for example, we talked about, you know, our chat GPT plugins, we talked about Ask Instacart, like it is a very obvious place to start putting some sponsored content so that if you're asking, you know, what are the like best dairy-free snacks for my kids, you know, we can show you organic results, but we can also show you some sponsored results there. And these are placements that are going to be really interesting for advertisers to tap into because again, there's a much clearer expression of intense and just a keyword and things that didn't exist before. So I'm really excited about that new wave and making that as relevant for people as possible.

26:21One of the things that people are talking about, I guess, related to both advertising, but also more generally in terms of e-commerce and shopping, is that the way we think about information delivery and transactions in the web over the next 5, 10 years may change dramatically. So basically, everybody is eventually represented by some form of agent. And you tell that agent, go and buy me these clothes or go and find these things for me. Go and do these things for me. How does that impact all this? How does that impact advertising? How does that impact commerce? Or is it so far in the future that it's not worth thinking too much about today?

26:56I actually think that it will have a lot of positive impact on commerce because again, first off, it's going to make people a lot more, either more productive or have more free time, which always leads to good things. A lot of these tasks are tasks that people would rather not be doing. There are some commerce tasks that they love, like browsing clothes, guilty as charged, or types of commerce that are mixed with entertainment. And I fully expect that for these, we're going to make things even more entertaining in the future through AI. And then for the things that are absolute trolls, these are the things that we're going to be able to automate a lot more.

27:43And so I think that's going to lead to really good results. The thing that has frustrated me the most with, in fact, Instacart is that it is an app that when you look at it today is fundamentally a utility. It's very transactional. You just like type keywords, add to a cart, it shows up at your door. It shouldn't be the case. Food is one of the most inspiring things. It's like it's one of the things that connects humans. It's one of the ways in which we celebrate, in which we show love. In fact, people spend more time watching food content than ever cooking it. And so, you know, that tells you that there's such an opportunity for like making this thing that everyone considers a chore of like grocery shopping, making it a fun moment where you discover a new recipe, where you get excited about what you're going to cook this week, where you discover a new snack that's going to be like the highlight of your week.