Enterprise AI Strategy: Bret Taylor on Lessons Learned at Salesforce

A TransformX Conference Highlight

At TransformX, we brought together a community of leaders, visionaries, practitioners, and researchers from several industries to explore the shift from research to reality within the realms of artificial intelligence (AI) and machine learning (ML).

Introducing Bret Taylor

Bret Taylor leads Salesforce’s global product vision, engineering, security, marketing, and communications. Taylor began his career by co-creating Google Maps, and he later co-founded FriendFeed, which was acquired by Facebook. He is credited with creating Facebook’s “like” button.

In this TransformX session, “Enterprise AI Strategy: Lessons Learned at Salesforce,” Taylor joined Scale AI CEO Alexandr Wang to discuss the key considerations that businesses of all sizes should keep in mind to ensure that they delight their customers and achieve sustainable business outcomes. You can also read an executive summary of the seven key takeaways here. The full transcript of the session appears below.

Alexandr Wang (00:22):

I'm excited to welcome our next speaker, Bret Taylor. As president and Chief Operating Officer Bret Taylor leads Salesforce's global product vision, engineering, security, marketing and communications. A respected software engineer and executive with an impressive history of building widely used and loved products. Bret was most recently the co-founder and CEO of Quibb, the innovative collaboration platform sales acquired in 2016. Prior to founding Quibb, Bret served as the chief technology officer of Facebook and saw the company through its successful IPO in 2012.

Alexandr Wang (00:57):

Credited with the invention of the like button, Bret joined Facebook in 2009 after he acquired his social networking company, Friend Feed. Prior to Friend Feed, Bret started his career at Google where he was the co-creator of Google Maps. Bret is a member of Twitter's board of directors, and he graduated from Stanford University in 2003 with a bachelor's and master's in computer science and now lives in the Bay Area.

Alexandr Wang (01:23):

Thank you so much for sitting down with us today, Bret. Always exciting to chat with you and excited to talk to you about AI and enterprises.

Bret Taylor (01:29):

Yeah, thanks for having me.

Alexandr Wang (01:31):

Yeah. So Salesforce has been a real leader in AI for enterprises and deploying AI in a bunch of different enterprise applications. It's obviously been great to watch the success of Einstein across all of your different business units and the businesses that you serve. So maybe just to get us started and to take a big step back, how did you first realize that AI was going to be an essential technology for enterprises?

Bret Taylor (01:59):

Well, especially post pandemic, there's so much to talk about digital transformation. I think it's because right now, if you don't have a digital business, you don't have a business, right? It's the only way to engage with your employees, your partners, your customers, your supply chain. I think every digital transformation is also a data transformation because the by-product of every digital interaction is more data. And the more data you can inform those digital interactions, the more intelligent and personalized they can be.

Bret Taylor (02:29):

So when you think about Salesforce does, we do digital sales, digital customer service, digital marketing, and digital commerce. In every single one of those areas, as the interactions get more and more digital and sales meetings move from conference rooms to Zooms and contact centers that used to be buildings are now just things in the cloud, right? All of that provides more data and more opportunity to use AI, to personalize those experiences, to make a better customer experience, to make a better employee experience. Just as an interesting example, unfortunately, there's a big surge in unemployment at the beginning of this pandemic and we have a lot of public sector customers, including the New Mexico department of Workforce Solutions. They got this overwhelming demand for people calling up for unemployment insurance. They deployed chatbots. Started doing about 7,500 chat bot conversations per day. People are able to solve their problems, get the unemployment solutions they needed and they didn't overwhelm their contact center. And that's just one example, but I think across our entire customer 360 platform, AI is central to everything that we do.

Alexandr Wang (03:39):

Yeah. You have a pretty interesting set of experiences in that you had worked at large-scale consumer internet companies. You were the CTO of Facebook, you're the inventor of Google Maps among other things, as well as now translating a lot of the experiences to the enterprise. What was your career wide kind of your first aha moment with AI in the first maybe like realization that this technology was going to be incredibly important?

Bret Taylor (04:06):

Yeah, it's interesting. I mean, certainly starting my career at Google, machine learning was a huge part of everything Google did. Doing a lot of it before it was cool. Before it was the hot topic in Silicon Valley for everything from spam filtering and Gmail to ranking search results. And this is prior to deep learning, a lot of the kind of technological approaches that are prevalent now. But the interesting thing by starting my career at Google, where we had too much data for any traditional algorithm to handle and so many consumers of our experiences that you had to scale things algorithmically, it created a culture that I think is very much the type of data culture that a lot of AI first companies have now.

Bret Taylor (04:55):

So I think it sort of set me off on the right foot in that way. I approach problems with that sense of scalability and where can we apply algorithmic intelligence to do things that our computers are better at than perhaps, or more scalable at than humans? So it had a deep impact on me. And I think Larry and Sergei always sort of imbued that in the culture and it stayed with me through my career, like most people's first jobs do.

Bret Taylor (05:21):

The interesting thing from applying those consumer lessons to Salesforce is that I think that there's that famous Marc Andreessen post, "Software's eating the world." And the premise of that post is that in a lot of markets, a software oriented startup will take a market share or an incumbent will learn how to do software and it's sort of a race just as digital experiences become the dominant way we engage with brands.

Bret Taylor (05:46):

It is such a privilege to work at Salesforce because we're helping companies that don't have buildings full of data scientists and buildings full of PhDs in machine learning to actually take advantage of these amazing technologies. For me, that's the next phase of these amazing technologies is making them accessible because if we really want all of our experiences to be smarter, more intelligent, more personalized, it can't just be in the domain of technology companies. I'm really excited and I feel a great privilege to partner with so many great brands around the world here at Salesforce.

Alexandr Wang (06:19):

Yeah. This point that you mentioned something that we talk a lot about at scale, which is like, first there's kind of this wave that's happening of software eating the world and that one by-product of software eating the world and software being used in all sorts of different applications is that all of a sudden you have digital data that is popping up. And as you mentioned, the pandemic has been a huge accelerant of this digital data, and that is one of the core enablers for any kind of AI or machine learning to empower your business.

Alexandr Wang (06:54):

One really big topic around AI, a lot of people in the community think about it. A lot of people when they think about how the technology can be used, it's the big topic is ethical use. What mechanisms have you put in place as you're thinking of Salesforce as a way to enable many other businesses and enterprises to use AI? What mechanisms do you put in place to ensure that AI is used ethically and it minimizes sort of any sort of bias or unintended consequences?

Bret Taylor (07:25):

It's a wonderful question. I think perhaps the most important question in AI right now. I think I've noticed through my career when I started my career in software and technology, there was perennial optimism. It's like, when is the whenever you get a new laptop and thanks to Moore's law will be 10 times faster. When is the next phone come out? I think the conversation has really changed as the whole economy has gone digital because so much of our life experience is based on these algorithms, these digital experiences that our relationship with technology has become more complex.

Bret Taylor (08:02):

We at Salesforce are more than anything, a values driven company. We start everything with our values, even with our mission. And our number one value is trust. I would argue that if you're building technology right now, trust has to be your number one value. If not trust, what is it? And you can see the negative ramifications when companies approach problems and trust isn't their number one value. It can lead to not just a degradation and trust in that product, but actually a degradation in trust in technology broadly. And you're seeing this, and if you sort of tear the headlines out of the Wall Street Journal, or you'll see just people's relationship with technology has really shifted. I think that the challenging question for all of us is not just having trust as your highest value, but how do you operationalize it? What does that actually mean?

Bret Taylor (08:50):

One thing we did at Salesforce is we actually named a chief ethical and humane use officer and create an office of humane and ethical use of technology. Paula Goldman who's serving in that role really serves as our advisor, bringing in outside experts into complex technology ethics decisions, but more than anything this may be before your time, Alex, but back when there was lots of security holes and desktop software, there was a principle that became popular called security by design, which essentially meant you have to design software with security in mind from day one.

Bret Taylor (09:27):

We have a principle at Salesforce, ethics by design, and in the product design process, we're thinking about ethical and humane use. We've released our ethics principles and from our research script and AI being AI should be responsible, accountable, transparent, empowering, inclusive. And we really are trying to bake that into our end-to-end product design process. And for any company that is applying artificial intelligence and machine learning, you have to take this into account from day one because you can't... Essentially these systems become the experience that you're providing to your employees, your partners, your customers. And if you don't take into account the design process, you will almost certainly get unintended negative consequences.

Alexandr Wang (10:12):

Yeah. Yeah. This is a good segue because I think right now we're at a point where every enterprise needs to develop an AI strategy. The technology is just, it's so important. It enables incredible things within a business. And so whether you're a startup or multinational enterprise like Salesforce or sort of an established company that's been in business for hundreds of years, it's really important to develop an AI strategy to sort of stay ahead.

Alexandr Wang (10:42):

I'm sure that Salesforce is a trusted advisor to many organizations, as well as you guys have developed your own AI strategy. What's some advice on building an AI strategy and how does that stack up or priorities against other goals within the business?

Bret Taylor (10:59):

Yeah, I agree with your premise. Every company needs to be kind of an AI company. And it's an imperative in the all digital work anywhere economy. Our experiences are going to be driven by these digital experiences and if your competitors provide more personalized, more automated experience, we'll have a better experience. So I agree with the premise.

Bret Taylor (11:20):

The scariest statistic, there was a 2021 Gartner's CIO survey that you've probably seen. So 94% of the CIOs who responded say they're working on AI projects. Only 24% of those CIOs had actually deployed one successfully. We are sort of in the peak of this hype cycle around AI. I would argue a lot of companies haven't actually seen the value they've hoped from some of those projects. It's admittedly somewhat biased because what we did at Salesforce, but my philosophy is don't start with the technology, start with your customers.

Bret Taylor (11:57):

Whoever the customer is, start with their experience and what you're trying to build. Don't build technology for technology's sake. And when you do that, and you start with the customer and you start with the principles sort of jobs to be done, what are you doing with this technology? It can really clarify the scope of these technology projects and help you find more success.

Bret Taylor (12:18):

The other thing I would say is that when I think about making AI more accessible is, find the easy wins. One of the things that sets Einstein, which is our AI platform apart is we're not building general purpose app platforms like you are at scale, we're building technology you can turn on with clicks, not code. I think there's a lot of hidden value in that. I mean, just as an example, we obviously integrate AI into our marketing cloud, which sends a lot of email and text messages and push notifications. One of our customers, Orvis was able to turn this on to

Bret Taylor (13:00):

... judged the engagement of the messages they sent out before they sent them out and they increased their click through rates by 22%. They increased their traffic by 30%. And it wasn't a really complex project. It was really just taking advantage of the tools embedded in these platforms, and thanks to companies like Scale I think a lot of these software and service tools are building intelligent capabilities into the products. I'm hopeful that that's a path for a lot of companies to get value from AI without becoming AI shops which is a really formidable challenge for a lot of companies that aren't native software companies right now.

Alexandr Wang (13:36):

Yeah. So, there exists a path. It isn't an all or nothing, hey, I have zero AI strategy or I have a team that's dedicated to building AI technology. There's this middle path which is I can work with partners that I've worked with for years and years and utilize some of the digital and AI technology that they bring to bear.

Bret Taylor (14:00):

[inaudible 00:14:00].

Alexandr Wang (14:01):

I'm curious, you touched on this a little bit. You mentioned this use case around the marketing cloud and you mentioned the chat bots a little bit ago. What are some of these use cases of AI that you've just seen tangible value amongst Salesforce's customers or just broadly within the enterprise? What are some of these that you encourage folks to be thinking more about?

Bret Taylor (14:23):

First, it's across the entire portfolio, because I've said if it's a digital experience it should be able to benefit from AI. And when I look at sales cloud one of the great examples of this is sales call coaching, and using essentially the ability to find mentions of competitors and calls. It gives sales managers the opportunity to at-scale provide coaching about how to make your sales teams more effective, especially as we're all effectively working from home with our headsets on.

Bret Taylor (14:55):

It's an amazing opportunity to drive sales performance. And our service cloud it goes without saying how relevant things like chat bots are, but one of my favorite examples is we have a product called Service Cloud Voice which is a contact center product and when you're on the phone where the customer's having a problem Einstein is transcribing what they're saying and suggesting the answer to their questions, which is really neat because wen you're on a support call what you want to do is get answered and get off the support call, both as the customer and as the company providing the support, and it's an amazing way of making service agents better, more productive, getting to that timed resolution faster than ever before.

Bret Taylor (15:36):

And in our B2C products like Marketing Cloud and Commerce Cloud it's all about personalization, and I think the bar has been set by digitally native companies like the Amazons of the world, and the Googles of the world, and the Apples of the world, and I think really what the promise of this is is you can have the same degree of personalization that those really sophisticated technology companies have.

Bret Taylor (16:00):

And as I said our aspiration, we don't always meet it, but our aspiration is you can just turn it on. And I think the most interesting thing when I see scale customers working with us too is the intersection between companies building proprietary models and systems and integrating them there with some softwares and services and I think we're still in the early innings of that.

Bret Taylor (16:23):

We have customers building custom models to predict attrition and integrating that with Sales Cloud. I still think we're figuring out that future but I see a ton of potential because as every company develops more sophisticated data science capabilities and they're adopting more softwares and service I think there's really great future opportunities to say how do these things all work together in a more seamless way.

Alexandr Wang (16:46):

Yeah. The ingredients are all there. The ingredients are do you have systems that are creating large pools of digital data. Thankfully, Salesforce among others as software and service providers have really created the preconditions for that to be the case, and it's huge. A lot of enterprises just as a byproduct of using products like Salesforce now have access to incredibly valuable data within their business.

Alexandr Wang (17:12):

Then it's like are there more and more people with machine learning expertise coming out? Is there more compute available? There's data availability. And you can combine all these together to produce incredible applications in the business, some of which are some of the most important problems that enterprises struggle with in the first place, as you mentioned, which is like how do you enable better more personalized experiences for your customers? That's a structural cross cutting issue.

Alexandr Wang (17:39):

One thing I'm curious about as you've seen it there's sort of a few kinds of different problems that AI can be used to solve with for these enterprises. There's customer facing problems, and that can be ... I would almost split it up between personalization is one aspect, but also enabling new customer experiences, like chat bots for example, or better voice assistance capability, et cetera.

Alexandr Wang (18:05):

And then, there's back office use cases. There's process optimization. There's making things work more efficiently internally. You all have seen a breadth of use cases. Where do you think it's most valuable for most enterprises to focus?

Bret Taylor (18:23):

Obviously, every business is different. So, if I was talking to any company with a large supply chain right now that's probably the top thing on their mind because of all the supply chain disruption and being more agile and intelligent around supply chain optimization could be the most important thing. However, what I would say is right now the focus of almost every CEO and every board that I talk to is growth.

Bret Taylor (18:48):

We just went through a major economic disruption with this pandemic. I think everyone's coming to the realization that this is not going to snap back. We're in a new pandemic world, and all digital work anywhere world, and companies are really saying how do we find our way back to growth in that new world. And fundamentally, that means building a new digital employee experience, a digital partner experience, and a digital customer experience.

Bret Taylor (19:14):

And I think just disproportionally, and maybe it's our vantage point in the enterprise software market, but disproportionally that's around customer experience, and because it can be the things that really change the top line growth trajectory of a company which is for most CEOs that's a pretty compelling area to focus on. And what I would say is this pandemic has really accelerated the change in expectations in consumers.

Bret Taylor (19:40):

Prior to this pandemic I had never shopped for groceries online. I'm more of the go in the grocery store and smell the fruit type of person. Now, I have a subscription to toilet paper. These things aren't going back. We've learned these experiences and we don't talk about Black Friday and Cyber Monday anymore. We just say Cyber Week. And no matter how many retail stores reopen the E-commerce is not going to go down next year. We've learned these habits.

Bret Taylor (20:07):

And so, I think focusing on the customer experience is probably the right place to start, and the right place to just get value from AI fast. The thing I would say though is going back to our first conversation, doing it way that's ethical, doing it in a way that's accessible, is really challenging, but I think it's worth it because every company needs to build intelligence into their business strategy.

Alexandr Wang (20:35):

Yeah, totally. And to your point I think as a level one approximation like we had mentioned just having capabilities built in almost automatically into your software and service platforms. That's a great way to just get on the AI track, get on the treadmill. But a lot of business, as you mentioned every business is different, they have unique challenges, they have unique problems where AI can be used in novel ways within their businesses, and so we like you have often seen sometimes enterprises struggle with going from experimentation to reality or experimentation to production.

Alexandr Wang (21:13):

How would you help enterprises think about how to actually build programs that will allow them to develop great machine learning internally? We mentioned it's a big hurdle, but you're one of the uniquely positioned people to answer this question because you're seen it happen at Google and Facebook and other internet companies, and have also just has a wealth of experience work with the traditional enterprises.

Bret Taylor (21:38):

It's a very challenging proposition for a lot of companies, and I think it's not unprecedented though. When the PC first came out the change management of adopting PCs in the workplace was very significant. It was a new way of working. When the internet connected and organizations were first adopting email it was a really dramatic shift in the way we work.

Bret Taylor (22:02):

The smartphone similarly, or I guess the BlackBerry if you're going way back in the enterprise, and then the smartphone revolution. And what I would say is machine learning's so unique because it's sort of a software problem, it's sort of a data problem, it's sort of as you know better than anyone in the world actually having your data be coherent while labeled high quality, having that pipeline reliable is as hard as any of the software problems. And as a constant just like these previous revolutions in the way we are working it's a new way of thinking about your technology processes and your business processes, and I think the best way to solve it is to bring on people who are experts and have done it before. And so, I think when you look ... I don't know when the first chief data officer appeared but it wasn't that long ago, and if you don't have one you should probably ask yourself why, right?

Bret Taylor (22:53):

And I think that there's roles that are being created at companies right now, and experts that you need to bring in to change the DNA of your company to really understand what this means. And I think the most challenging thing is AI is just hard enough right now that it's hard for a lay person to intuit which applications will be effectively solved through a current state of machine learning, and if you don't bring on the right expertise you can end up one of those [inaudible 00:23:22] of the failed projects.

Bret Taylor (23:23):

So, I think it's really about culture. It's really bringing on that expertise. It's also recognizing the gravity of the change management that you'll need to go through if you really want to become a data first company.

Alexandr Wang (23:34):

Yeah. And one of the things I think from my experience speaking with enterprise that I think is a question or a struggle is what's the right investment model in the sense that it's great if there's cases where you can turn it on in some existing solution like Salesforce and that's great obviously, but in starting these AI programs or investing into AI programs you're right we have been in a phase where a lot of experiments have failed which causes it to be a tricky thing to feel good about continuing to invest a lot of resources into. At the same time, the potential upside and the value of getting AI working in your business is so, so great.

Alexandr Wang (24:17):

So, how do you think about the right investment framework or philosophy around investing in the business case of AI within the enterprise?

Bret Taylor (24:27):

I think the main thing is to understand why you're building what you're building and I say all start with the customer. Whether it's an internal customer or an external know the impact you're going for and if you have a clear business case, you have clear business value, and the funding becomes obvious.

Bret Taylor (24:44):

When I think about for example our Einstein built into commerce cloud the measure's really clear for our customers. They want more gross merchandising volume running through their system, and Icebreaker which is an outdoor apparel company when they turned it on they basically saw recommendations clicked on 40% more often and 48% more revenue from recommended products. Very easy to measure, right?

Bret Taylor (25:10):

Unfortunately, not all projects have a dollar sign on the other end like that, but if you spend a lot of time recognizing what the business value is you're trying to achieve it becomes really easy to fund it because you know once the project succeeds here's the impact it can have on either a cost savings or top line growth and it can be a really clarifying thing so the project doesn't get away from you.

Bret Taylor (25:35):

I've just seen a lot of people where the AI becomes the purpose of the project and you end up with a bottomless pit of investment and also a really unclear definition of success and a definition of done and I think that's what smart CTOs and smart CIOs are being very opinionated about. Here's the outcomes we're trying to deliver.

Alexandr Wang (25:55):

So, we talked a little bit about the talent hurdles. We mentioned a little bit that getting data in the right places

Alexandr Wang (26:01):

... is critical for these enterprises. What are the top recommendations, let's say you're a CEO of an enterprise and you take this recommendation, hey, we're going to be focused on what end metrics are we trying to affect with AI, what is almost the checklist or the key fundamentals to get in place before being able to build great... to even get started on the problem?

Bret Taylor (26:28):

One of the things that I believe in is being practical. And I've met a lot of executives who there's a little bit of the science fiction view of AI, and it's like you turn it on and everything just works in an automated way. I think there's so many short-term opportunities to just make all of us as employees more productive with a little help from AI.

Bret Taylor (26:54):

I'll give you a really funny example of this. So we have a capability in our sales file called Einstein Forecasting, and essentially it helps you predict how many sales you're going to close in the quarter. When we run our sales forecast call, we have the roll-up from our sales leaders, and they say, "Here's what we think we're going to hit this quarter based on our opportunities and the pipeline," and we have Einstein's number right next to it. So it's like here's the sales leader's number and here's Einstein's number.

Bret Taylor (27:24):

So one time, Einstein thought that one of our sales leaders was going to miss his number, and he was furious. He was like, "Einstein doesn't know what he's talking about." And it turns out he was right. What was really going on is there was a data problem. So some of his sales reps weren't properly updating Salesforce after some of their customer calls. Qualitatively, he knew that some of those deals were more likely to close, but the data wasn't reflected in Salesforce.

Bret Taylor (27:57):

And what was interesting about it, though, is that sales executive became the strongest advocate for Einstein because even though Einstein had an incorrect prediction, it actually helped him fix an operational problem on his team. People weren't actually updating the single source of truth that was Salesforce. And he said, "Look, if the data is not in Salesforce, it doesn't exist."

Bret Taylor (28:19):

And he actually created new rigor on his team to make sure the data was already up-to-date. And he tells that story now to our customers, talking about how this actually helped him uncover systemic human issues on his team, even though the prediction was incorrect. And I bring that up just because I do think it's a vision for how much we can use intelligence, even if it's imperfect, to help us become more effective as organizations.

Bret Taylor (28:45):

And so, I would just say that you don't need science fiction to get value from AI. And I think that's, when you scale it back and you say, "What wins can we generate right now?" Often, there's a lot of relatively low-hanging fruit right in front of you.

Alexandr Wang (29:00):

Yeah. Well, I've noticed in general you've really been quite focused on, hey, there's low-hanging fruit that exists within the business where there's a lot of short term value propositions that exist from implementing AI. On a slightly longer time horizon, what do you think the sort of enterprise of the future... Let's think an optimal enterprise for the future, what does that look like? What does AI actually enable for that business, let's say five or 10 years out?

Bret Taylor (29:32):

Yeah. I think the biggest opportunity I see is really getting better about predictive intelligence. I think that when we get to a point where these systems of record at your company can actually tell you things you didn't foresee, I think it will really change your relationship with the software that you're using. And I think that, especially with Tableau and Einstein and some of the investments we've made here, we're still relying a lot on the analyst and executive sort of saying, "Here's the prediction we want to make."

Bret Taylor (30:11):

And I do think that as systems get more intelligent and can understand more of the context of your business automatically, I think there's an amazing opportunity to look around corners that executive teams and boardrooms didn't necessarily know to look around. And I think that can be a really powerful thing for businesses.

Bret Taylor (30:31):

And I think there's pockets of this today in kind of specialized verticals, but I do think there's an opportunity to really generalize this. And I think it will make executive teams look stronger than they ever were because it's essentially making your accountability and predictability of your business even greater.

Alexandr Wang (30:51):

Yep. One of the things that sets Salesforce apart a little bit from many other enterprise software companies is that you have a vibrant AI research effort and a vibrant AI research lab. What were some of the most interesting developments or research that have come out of Salesforce over the past few years, and how do you think those will ultimately translate into the way that enterprises use AI?

Bret Taylor (31:17):

Well, first, just so grateful to our research team. I think that, as you all know and as all of the viewers of this conference know, the pace of innovation and deep learning, machine learning, and natural language processing is so rapid that there's kind of a blurry line between research and application.

Bret Taylor (31:36):

And so, I think it's really important for us as an organization to not just focus on what is applicable today, but to have groups looking beyond the horizon, not necessarily focused on commercial viability, but just saying, "How can we push the envelope with this technology?" And it's become such an asset for us to, in a slightly better way, actually predict the future of the applications as well. So a lot of research around natural language processing, conversational AI.

Bret Taylor (32:06):

One really fun example, Marc Benioff, as many know, is very passionate about the oceans, and particularly getting plastic out of the oceans and kind of coexisting with nature in a better way. We partnered with UC Santa Barbara for a service called Shark Guide to actually find great white sharks and help get people off the beaches in California when there are great white sharks in the ocean. And it was a really fun project.

Bret Taylor (32:32):

But what's really great about our research group too is they have access to our customers. So they've been able to take a lot of their research ideas and test them in the field with customers. For him, it's a business critical problem and a really fun place to do research.

Alexandr Wang (32:48):

Yeah. Awesome. I want to go back to something that you had mentioned before, which is you had this, you're very grateful that you had sort of started your career at Google. And so you sort of were steeped in this almost like data-centric, algorithmically-centric, scalability-centric culture. What do you think are the core tenants of that? What are the reflections or the sort of principles of a data-driven culture versus one that isn't? And how would you start to instill that with an organization?

Bret Taylor (33:22):

Yeah, I mean, it's interesting you mentioned the phrase data culture. It's actually probably the key tenant of Tableau, which is our data analytics product, is to create data cultures and help everyone at your organization see and understand data.

Bret Taylor (33:38):

I think that word culture is the most important thing. As much as I advocated for people having a chief data officer, if data and intelligence is in a department at your company and doesn't break out of that department, you're not really a data culture.

Bret Taylor (33:52):

What a data culture means is that everyone is making data-informed decisions and their decisions aren't made on their gut instinct. They're not based on who the highest paid person in the room is. It's based on actual data and objective view of your customers, of your supply chain, of your employees.

Bret Taylor (34:09):

And I think a big part of that, one of the principles, is recognizing it's a cultural shift and figuring out where you're making decisions, and data is at the starting point. And a lot of that is change management and operational more than it's technical.

Bret Taylor (34:24):

The second thing is the tools you give your employees. One of the reasons we were so excited to acquire Tableau is just how much it was focused on empowerment. It's about as easy to use as Microsoft Excel, but it's a super powerful tool. And it meant that being an analyst wasn't the domain of experts, it was a domain of everybody.

Bret Taylor (34:46):

And I think that's a really important next step for AI and creating these data cultures is making sure that these systems aren't so rigid that there's a data scientist in some building over here making all the data decisions for your company, how do you actually distribute that accountability and distribute that and make it empowering for everyone at your organization?

Bret Taylor (35:07):

And I think naturally companies with data cultures will win. And so, I think, I know in a decade most companies will have data cultures because in a decade the companies that created them will have outgrown their competitors.

Alexandr Wang (35:21):

Yeah. And when you mention data... You touched on one important aspect, which is using data to drive human decision-making within the company. And then there's kind of this next level of almost what I would call the algorithmic data culture, the AI inspired data culture, where, hey, the culture is that you are stewards of data within your business that are then fed into algorithms and those algorithms then also make decisions in your business. And so, you're removing people from the sort of decision-making process.

Alexandr Wang (35:52):

How do you think about enabling that as much as possible within an enterprise? How do you think about that transition for most enterprises to move from a predominantly human intuition based decision-making culture to one where you're trusting algorithms to make a lot of decisions within the business?

Bret Taylor (36:09):

Yeah. Automation is the top priority of every CIO and CTO I talk to. And it's in part because if you look around most enterprises right now, there are so many areas where people are copying and pasting between tools, using spreadsheets and PowerPoint presentations to do things that could be entirely automated.

Bret Taylor (36:30):

And actually, I'll just give you a really interesting example from our own business. We have a customer experience team. We do over 10,000 surveys a year just to kind of measure the sentiment of our customers. It used to take, as you can imagine, like 15 hours to comb through like thousands of surveys. We now do almost all of this in an automated way, and essentially doing sentiment analysis and other things so that our team can actually scale up the number of surveys we're sending and get more value from it by just removing almost all manual processes from the loop.

Bret Taylor (37:06):

And I do think, as you described, it's kind of the next level of applying data and intelligence. I think both matter a lot. I think there are things that should be completely automated. And I also think that AI, machine learning, analytics can help all the decision-makers at your company have better intuition as well. And I think it's probably incorrect to think about it as one or the other. I think both go together when you're thinking about creating a data culture.

Alexandr Wang (37:34):

I think, to kind of go back to it, I think part, in building AI into, into a lot of businesses, definitely one vector that is on a lot of CEO's minds is, okay, we want to use AI to reignite growth. In many ways, that's going to be through enabling new user experiences, right? And this is something that traditionally internet tech companies, this was like the key thing that they were able to do is innovate very quickly on new ways to new user experiences that then drove greater engagement or drove greater commerce, et cetera.

Alexandr Wang (38:09):

What recommendations would you make, just given your own experience literally building a number of user experiences that are critical to the average person's life today, whether that be Google Maps or at Facebook or Gmail or whatnot? What is your advice to enterprises who want to be innovating on how they interact with their customers and what end user experiences they enable while still staying true to their sort of core brand identity?

Bret Taylor (38:41):

Well, it's going to sound funny, but if you want to build a better customer experience, focus on your customers. And what's interesting about it is a lot of companies talk about customer experience, but if you look at the way they're organized, they're organized around their products. They're organized by their regions. How many times have

Bret Taylor (39:00):

I had an experience where you have maybe a conglomerate that's done a number of acquisitions and they don't have a single view of you as their customer, the right hand is not talking to the left hand.

Bret Taylor (39:10):

I think the most important transformation that every company in the world needs to make is organize themselves around their customers, whether it's a bank and maybe you have a bank account over here, and an investment banking relationship over here, maybe you've started a small business and gotten a loan, you want a single view of you, Alex, as the banking customer, right?

Bret Taylor (39:31):

And most companies just miss on that. And we have this idea at Salesforce called Customer 360. And it's really about saying, "Can you have a single view of your customer across those touch points?" The basis of that is having a single source of truth about your customer data.

Bret Taylor (39:46):

And if you have that, building automated and intelligent experiences on top of that is available to you. If you don't and your data is siloed, you don't, and that's the blunt truth. And I actually think that it all starts with the data and it all starts with, are you actually putting your customer at the center of your business or are you shipping your org chart? And I think it's one of the biggest challenges most companies face and a huge opportunity for customer-centric companies.

Alexandr Wang (40:10):

Yeah, totally. And this is something that we see all the time with our customers, is that our whole thesis is that data is the foundation of being able to build any sort of reasonable AI strategy, and the exact point that you mentioned, data is siloed within most enterprises, and pieces of data that are very relevant to any sort of application or intelligent experience that you'd want to build on top of it, they're split out based on the org chart. I think what you just mentioned is a great soundbite.

Alexandr Wang (40:38):

You know, we've been talking through all sorts of different applications of AI within enterprises. In many ways these are very tangible applications, they don't necessarily mirror the sexy things that you see on the internet about all the various ways that AI is painting paintings, or folding proteins or whatnot, and all these innovative new applications.

Alexandr Wang (41:01):

But one implicit thing that I think you and I can both agree on is the time like, and the question is if you agree with this, is like the time to invest is now, like the time to be investing into what your business will look like in an AI-enhanced way is now, and you shouldn't be waiting for some magical discovery in the future.

Bret Taylor (41:24):

I agree. It's interesting, there's a great phrase I think often credited to Bill Gates, which is, "We often overestimate what we can do in a year, and underestimate what we can do in a decade."

Bret Taylor (41:36):

And I think that particularly when you have technologies like deep learning, and other forms of machine learning that are accelerating at such an exponential pace, a decade from now I can assure you, a lot of these sort of science fiction ideas will be a reality, at least a meaningful subset of them.

Bret Taylor (41:54):

But if you're not making those foundational investments now, creating a data culture at your company, building a digital customer experience, building a single source of truth, breaking down those data silos now, you'll be caught flat-footed once the technologies are actually a reality.

Bret Taylor (42:10):

And if you need any proof of that, look at the companies that went into 2020 and the pandemic, and how you felt when the companies that had invested in digital technologies at least had a business then, and the companies that were building on the fly lost six months of business.

Bret Taylor (42:28):

And I think that this transformation of our economy to turn digital is accelerating. And really, I think there's no shortage of the investment a smart management team should make in digital customer experiences, AI, and intelligence, and personalization to set themselves up to grow in that all-digital, work-anymore world.

Alexandr Wang (42:49):

Yeah. And then kind of in this vein, to your point, like, yeah, I love this quote that we overestimate what we can do in a year, underestimate what we can do in a decade. Where do you believe that AI is going to have the most significant impacts? And this is just globally, not necessarily just constrained to impact to the enterprise, the most significant impact over the next five years.

Bret Taylor (43:16):

Oh gosh, that's a question maybe beyond my pay grade, because it's so pervasive. I was interested in one of the most interesting articles well outside of my area of expertise, was reading about how the mRNA vaccines were developed largely through software. You hear a lot about cancer diagnosis and the impact on healthcare. You look at the financial services industry and the opportunity to reduce fraud and improve the customer experience. We've talked a lot about retail, and just how amazing personalization is to our day-to-day consumer experiences now. And then I talk about the nuts and bolts of things like removing spam from social media platforms, email inboxes.

Bret Taylor (44:05):

Across the board I think what's really powerful about the development of machine learning technologies is it should, if we do it well, make every single one of our consumer experiences better. And when we look back, obviously things like our health are of paramount importance, maybe that will be the biggest impact.

Bret Taylor (44:27):

But I think in general if all of us as technology companies and all the companies that are partnering with Salesforce and Scale and others, building more intelligent experiences is just, it will take friction out of our lives. Customer experiences will be better, our day-to-day experiences will be better, smarter, more productive.

Bret Taylor (44:47):

And I think that's a really wonderful opportunity, because when you think about why does technology exist, it's to make our lives better. And I think the automation afforded by AI is one of the most promising developments in technology, in the technology history.

Alexandr Wang (44:59):

Yeah. And then maybe one last question is like, I think you have a unique view, you've seen, as you mentioned, you were in sort of the period of great optimism of technology where, "Hey, technology is just purely making everything better." And then as many of these technology platforms scaled up, we saw the potential unintended consequences of that technology.

Alexandr Wang (45:26):

What call to action, or what recommendation would you give to people developing AI technology, many of whom are attending the conference today, in really ensuring that we can have a positive future of AI, that there aren't as many unintended consequences as there could be if we weren't focused on it?

Bret Taylor (45:45):

Well, I think the first thing to recognize is that technology is not inherently good or bad, it's how you use it. And I think the days of assuming any technological advancement is purely good is, I think, rightfully being characterized as naive.

Bret Taylor (46:03):

And it's interesting, I think that in most other disciplines ethics are a huge part of what you learn if you're becoming a physicist or a chemist, and you learn about the history of applications of those disciplines that didn't benefit humanity. And it's a big part of the discourse of new development.

Bret Taylor (46:27):

And I think with the headlines around technology, misinformation, bias embedded in machine learning models, whatever it may be, I think society is now holding software and artificial intelligence to a higher standard. And I think all of the technologists watching right now who are developing AI embrace that, embed conversations about ethics and trust into your development process.

Bret Taylor (46:59):

And it probably involves bringing stakeholders in who aren't just data scientists who can be a part of the design process and look around the corner.

Bret Taylor (47:07):

I think the thing that is no longer acceptable is you didn't have the right person in the room, you created a technology that had a massively bad unintended consequence, not having those people in the room is no longer an excuse.

Bret Taylor (47:21):

And so I think we really need to bring in more stakeholders into the technology design process, ethicists, sociologists, and others who can help us shape these technologies in a more responsible way. And I think that's going to really change the way we think about technology development, and I would argue it will change it much more for the better.

Alexandr Wang (47:40):

Yeah. And maybe just a last message to kind of close. I think that for many enterprises this journey of AI is going to be incredibly exciting, but one that's incredibly daunting, one with potentially many challenges. What would your advice be to just like... What should they keep in mind as they go on this sort of multi-year journey? How can they ensure that even with bumps in the road that they'll ultimately end up in a state where the enterprise is AI-imbued?

Bret Taylor (48:11):

Well, I think the consumer expectations around customer experience now have permanently shifted. And so whether or not you want it, you're being compared to all the digitally native companies right now. And the answer is to build a Customer 360, and apply AI to build personalized, more automated, faster, easier experiences for your customers.

Bret Taylor (48:38):

And I think that what's really amazing about the digitization of the economy this past year is I think it's given permission to try bolder things, and from the boardroom on down, because if you don't have a digital business, you don't have a business.

Bret Taylor (48:55):

So don't let the crisis go to waste, use the opportunity of this, just seismic shift in expectations and technology, and use it as an opportunity to make a bold investment. And on the other side of it, I think you have the opportunity to gain market share, gain customer loyalty, gain employee loyalty, and right now there's never been a better opportunity to invest.

Alexandr Wang (49:20):

Brett, thank you so much again for sitting down and joining TransformX, it was a real pleasure.

Bret Taylor (49:25):

Thanks for having me.