Enterprise GTM Leadership
I was attending SaaStr this year, and I went to a session, and it had Tomás Tungás as the moderator, and I saw our next guest, Maggie Hott, and I'm like, we've got to get her at SaaS Metrics Palooza. And with a little bit of luck and blessing from someone above, we got Maggie to be able to do this. So let me introduce Maggie Hott, GTM leadership at OpenAI. Hey, Maggie. Hi. Great to be here. Thanks for having me. Okay. It's your stage. All right. Let's take it away. Hi, y'all. I'm Maggie. I'm on the go-to-market leadership team at OpenAI. I specifically lead all things go-to-market for chat GPT enterprise, as well as our account associate function, which is basically our demand gen motion. We're going to talk about three different things today. First off, we're going to start off with OpenAI's journey. How did we get here today? What is chat GPT, chat GPT enterprise? And then finally, we're going to wrap it up with some insights into the future of AI. So what is really this next wave of frontier, and what can we expect for the roads ahead? All right. So I always like to start off with this, of a little reminder of how long we've really been in kind of this AI excitement hype train that is going on. It's really only been since November 30th of last year. And this is an internal note that came out to us that essentially said, hi, we're launching chat GPT today. We're doing a silent launch at 10 a.m. Don't expect any significant impact on sales, on audiences. It's really just mostly researchers. And really the goal of releasing chat GPT to the world was to be able to get our models out so we could experiment, so we could start to get feedback. So the silent launch really over Thanksgiving week, leading up to the holidays, really not the time that you ever want to launch a product, is when we launched it. And all of a sudden, something started happening. We could tell within a few hours, the world started picking up on this, and it was going viral. All of a sudden, people were using chat GPT, people were talking about it, and we were completely blown away by the support and by the demand. It has been said that we are one of the fastest growing apps or consumer apps to ever hit the world. As you can see here, there's been speculation out there that we reached over 100 million users just two months after our launch, which is pretty phenomenal, again, for a product that really wasn't marketed or publicized or really meant to go in and kind of take the scene like that. So a little bit about our sales team. We went from having about 30 leads per week to 10,000 with only three sales reps. So how did we get here? How did we get to this moment? Chat GPT was not just developed overnight, even though it may seem like that to some. OpenAI actually started as a research lab in 2015. Our goal is to build safe and responsible AGI, artificial general intelligence that really benefits all of humanity. Think of a world where every single person has access to help them navigate a complex healthcare problem, or every single student, no matter where you live, now has your own personalized AI tutor. Those are some of the goals and some of the missions that we hope to achieve. So in June 11th, 2020, we released our research as an API. We started with four models, ranging from Ada being the most simple one where you could do classification to DaVinci, which was actually quite complex, and we were doing generation, and we started seeing companies like Jasper and Sana Labs actually infusing this into their products, or in those cases, building entire companies off of these models. Since 2021, we have really picked up steam. I'm not going to walk you through every single product on here, because as you can see, it's a lot, but there's a couple in particular that I want to call out that were very monumental. One of the ones here being DALI 2. It has just been a little over a year that DALI has been out, but funny enough, it kind of feels like an old hat trick at this point. Seems like DALI is everywhere, image generation is everywhere. This has been a pretty short time, and we just two weeks ago announced to the world that we have built DALI 3, which I'll give you a sneak preview of later on. We released ChatGPT, as we talked about last November, ChatGPT+, this last February, started creating a bunch of new models and APIs with the Whisper and ChatGPT API, GPT-4, function calling, iOS apps, Android apps. This company has been coming out with our research and our innovations at a faster clip than I have ever seen any other company do, and it is a ton of fun. Something else that I want to touch on is the power of GPT-4, which is really what is powering ChatGPT Enterprise, as well as ChatGPT+, as well as it's a model that is publicly available now to any company in the world. GPT-4 is likely the most complex piece of software humanity has ever produced. That is a very big and very bold statement, but the work and the output that GPT-4 can do is unmatched to any other model that is out there. And don't just take our word for it. There was actually a study done, which you all have heard about, from BCG Consultants who ran a test of consultants using AI versus not using AI. So essentially they took over 700 consultants. We did not partner with them on this. They broke them out into three groups, those who didn't use AI, those who used AI, and those who used AI with training. You'll see the results here over on the left-hand side, but there's two things I really want to call out. One, they were able to complete their tasks 25% more quickly. Imagine what your life would be like if you had 25% of your day back for leisure, to get more work done, to be with your family, and you were spending all that less time on administrative tasks. It also produced 40% higher quality results than those who didn't use AI within their work streams. Imagine if your work output was 40% better than it is today. So we started seeing companies shipping real products on and with our API. They built chatbots that serve both fun and functional purposes. If you look at HubSpot, they built ChatSpot, which allows you as the end user to interact with your data in a very natural language way. We had Discord build Clyde. Snapchat took the world by storm with MyAI, which the Gen Zers totally love. I'm not sure I fully use it, but it has taken the world by storm to help give you coaching on how to make the best snaps. There's been some world-class recommendation engines that have come out with this. Spotify has done its own personal DJ and just two weeks ago launched that now you can listen to some of your favorite podcasts in multiple different languages in the narrator's voice. So if you want to hear Lex Friedman in Spanish, you can absolutely do that. We've got Instacart, who is reimagining grocery delivery and grocery shopping, where you can input in what is in your fridge and it will create custom recipes for you and add it to your cart. Having to take away all of this, as a busy mom of two, I spend a lot of time meal planning. This helps to take away so much of that and automates my life. There's been tools built to catalyze users. AstraZeneca is reducing time and cost associated with drug discovery by leveraging Notions AI. And Ironclad is helping legal teams move a lot faster, which directly benefits sales teams when they are able to close deals and leverage the AI built within Ironclad. So there has been, this has all been pretty much over the last year since ChatGPT came out where it really opened up the eyes of the world to the power of AI. And you can see some of our customers here that we are so fortunate to work with that have been infusing GPT-4 and our technologies into their products. But while all this was going on, we were still noticing behind the scenes that 80% of the Fortune 500 was starting to adopt ChatGPT, free, and then as of February, ChatGPT+. This is shadow IT here, essentially. Customers or these end users are going, they're leveraging ChatGPT for their work, they're inputting in analysis, we'll get through some of the use cases here. But we realized while we were focusing on our APIs, that there was this aha moment of there is something here that is also for the enterprise, because this type of enterprise adoption doesn't happen unless people find something very valuable. So last month we launched ChatGPT Enterprise, and this is what my, the team that I oversell here at the Go-To-Market team. We have now three different products, ChatGPT Consumer for free and plus we've got our API as well as ChatGPT Enterprise. Our API and ChatGPT Enterprise are our true enterprise grade products. We will never train on your data, they are fast, they are reliable, and we'll get into some of the other details with them in just a minute. Just a quick snapshot of, again, on August 28th, kind of felt like we took the world by storm again by announcing ChatGPT Enterprise. But something I want to share, which was really important to how we built this product, is the partners that we worked with in development. We spent many months prior to the launch working with all of these amazing companies to shape what it was that they needed to feel comfortable with a V1 model of this product. So we're looking at companies like Amgen and PwC, to Sourcegraph with code generation, to Riot Games, to Asana and Bain. And we wanted to make sure that when we came out with the V1 of this product, that it would have perfect product market fit. So let's talk a little bit about what ChatGPT Enterprise is and how, in just the last month and a half, it has been impacting users. ChatGPT Plus is phenomenal. I assume many of you have used it before. It's got GPT-4, it's got Advanced Data Analytics, which was formerly named Code Interpreter, and it's got plugins. In its own, it's a really robust, really incredible package. However, I have spent my entire career in PLG, and one of the things that PLG companies do wrong is they don't think enough about feature differentiation between a freemium product and an enterprise-grade product. And then you have these buyers saying, why should I spend 2x, 3x, 10x, 50x for your enterprise product? We took that principle into mind as we were thinking about developing this, and we decided, let's come out with something that is so incredibly differentiated that is going to catapult how every single knowledge worker does their work. So this is what we built. We went from just this over to here. We loaded it up with security and access controls. We know there's nothing that is more important to a large enterprise than having data privacy assurance, things like custom data retention, SAML-based SSO, SCIM, RBAC, all the things that are all the acronyms that companies really, really need. We built insights and reporting. People want to know how their employees are using this tool and how often. We built template libraries, which is one of my favorite features I'll talk to you about in a minute. We are right now building in custom data sources, basically the ability to have your data infused with ChatGPT and for ChatGPT to be able to go across all of these different systems and services and pull in things that are tailored to you and your company and your role. And finally, we elevated the performance of this. It is fast, fast, fast, uncapped, unlimited GPT-4, which sounds silly, but literally, you can't get this anywhere else. Finally, we wrapped a bow on this with customer success. It is really important that we give a white glove service to these enterprise accounts that are trusting us with their entire workforce and how they're doing their day-to-day jobs. So we are rolling out this red carpet for them, and we are working hand-in-hand with these enterprises to help enable all of their employees to better use AI. And then finally, of course, last but not least, is we sprinted out the security certifications, SOC2, CPA, Encryption at Rest, Transit, all of the fun acronyms. Actually, in fact, when we launched, we were already in the audit period for SOC2 Type 2, which is pretty unheard of for day one to already be entering into that audit period. So let's talk a little bit about the use cases here. Cat GPT Enterprise isn't just an assistant. It's not just a chat box where you can ask it simple questions. Cat GPT Enterprise is here at OpenAI. This is our internal productivity tool. We do everything within this tool, and every single employee now has an individual chief of staff. Imagine if each one of your SDRs had a chief of staff, each one of your managers, even the people who manage your facilities can have their own chiefs of staff. Think about how much more productive that is going to make every single one of your employees. So let's actually break down a real world use case on how we cut our territories internally at OpenAI. Y'all know, most of you are in sales and marketing, cutting of territories every year is one of the most painful things that ever happens. Up until OpenAI, I would spend every single Christmas break, prior to this, I worked at Slack and at Webflow, and we would be spending hours and hours on Christmas Eve in the data, trying to make clear, equitable, balanced territories so that we could have them ready for the launch of the fiscal on February 1st. So we would always start this process months in advance. Now within chat GPT, using our advanced data analytics, this can be done in minutes. You probably wouldn't just want to put out the final version right away, but think about how you do things like balance a territory. You export tons of Salesforce reports, you export internal data usage and metrics. You're looking at things from all different systems and tools. And now imagine being able to upload all of those files all together in one and having it analyze and create all of these new territories for you. And this is just one small example that typically takes teams of folks from rev ops and sales leaderships weeks to do. You can also understand your reps' performance. So as we know, reps' performance and trends sit within many tools. They might be in Salesforce, it might be in Gong, it might be in outreach, and now you can have one single pane of glass where you can take all of the data across your entire company and you can really understand who is performing, what are trends, what are their win rates. We actually used to, at Slack, we had someone that was nearly full time who would do this, who would sit and tie together all the different systems and services and tools and produce a monthly report of what all the reps' performance was. This can again now be done in minutes and you could create a repeatable workflow that you could run this every week, every month, or however often you wanted. You can up-level your coaching. Imagine everyone now has their own dedicated one-on-one manager to go to, to practice with, to run scenarios by. Something that I'll do with my team is we will download transcripts from our team meetings or from one-on-ones and it'll give us tips and tricks on how we could have had a more impactful pitch or a more impactful team meeting or a more impactful coaching session, you know, if we're doing discovery role plays, what are some things that we could have asked that we didn't. It is insanely powerful. So ultimately what you're doing is you are turning every single one of your reps and every single one of your marketers and every single one of your employees into an AI power user with this tool. And one of the best ways to do that is with templates. We also launched templates within ChatGPT Enterprise. I will take full credit that this is not, this was not our idea. This came from our pilot users. This is how ingrained they were in our development process is we were finding our pilot users were creating Google Docs and storing all of their templates and then copying and pasting it and sharing it across the company and it was really messy and we said there had to be a better way with this. So we basically created this template directory where anyone in the company can take their superpowers within ChatGPT Enterprise and share it across the company. So imagine that one SDR that always has the best outbound emails. You can now replicate them across your entire sales dev force. That is so powerful. Couple other things on advanced data analysis because I believe this is probably one of the most powerful tools in the world right now. You can analyze and you can visualize data. You can upload in virtually any type of report from PDF to Excel to JSON to whatever you want to put in here and it can read it and it can understand it and it can create charts and graphs. Imagine all of those hours that you all have spent building presentations over the years. That completely goes away now with this. Here's just more examples. I'll click through them pretty quickly, but this tool can pretty much do anything. It is essentially Python running under the hood. It will spin up a Jupyter notebook and you can always see all of the work that is being done for every single one of these different queries. I want to give you all a look into a top five FinSERF, can't say the name, but basically what they recently did is they were looking across about 50 different companies, a mixture of private, of public. They had internally confidential data. They were using external publicly available data, 10Ks, and they needed to do a lot of analysis to understand the positionings of these companies and what might be a good investment to make. They were able to use chat GPT enterprise, in particular advanced data analysis, to run the complex Monte Carlo simulations, to do things that they could have never done before in Excel, and they were able to take their time of their analysts from weeks down to one or should I take what used to be many weeks down to minutes, which is pretty wild. So let's chat a bit about what does the future hold? Where are we going? What is next? You now know about our APIs, you know about chat GPT enterprise, and you know just a couple of the use cases that are transforming how people run their day to day. I believe the future is multimodal. To date, we have largely been very text based, text in, text out, but as we announced two weeks ago, chat GPT can now see, hear, and speak. This is pretty wild. Just let that digest for a second. You can talk to it, you can listen to it, you can see it, you can visualize it, and we'll take a look at what some of this looks like. But really the takeaway here is that the next wave of AI is multimodal. Let's start off with speech. We announced this two weeks ago, and this is one of my absolute favorite functionalities, tell you why in a second. But essentially, you can engage back and forth with chat GPT. You're on a long car ride. Tell me in 23 minutes about whatever it is that I want to learn about. You want someone to help you learn Spanish. Okay, quiz me back and forth for 15 minutes, ask me what it is that I can say, and virtually anything that you want to know that you would have been able to get out of the power of chat GPT, you can now do it verbally. For me personally, I've got two little girls, a one-and-a-half-year-old and a three-and-a-half-year-old, and my three-and-a-half-year-old doesn't realize that she has been playing with some of the most powerful AI technology in the world. Every single night, she's very into Ariel and Elsa right now, A Little Mermaid and Frozen, for those of you that might not have kids. Every single night, we talk to this and it tells us stories. She incorporates in, Logan, tell me a story about Logan and Ariel riding a unicorn on the beach, drinking a green smoothie. I mean, you can give this thing anything, and all of a sudden, it just lights her up and lights up her imagination to have someone tell her stories exactly what she wants to listen and learn and think about. Chat GPT vision is unbelievable. The short of it here is that you can upload in or take a picture of anything in the world. Anything that you want, you can ask it what the picture is, you can ask it to tell you in this case, tell me about the rules of baseball, create a chart, show a table, everything based on this picture. It is actually smart enough to know when there's an error in the picture, and it'll actually call out what the errors are in the image. Think if you were someone that is working for Insight and just start uploading in all of the images of your products, and it could write you descriptions, it could do meta-tagging. If you maybe are doing, for example, like holiday products, you could tell it, have it to do in the voice of Santa with holiday jingles. This thing is so powerful, and this will be first released in Chat GPT Enterprise coming very soon. As well as Dolly, going back to my daughter's imagination, this is also something that we play with every day. She called it making photos, but Dolly, you can tell it very simplistically. Create an image of a girl holding up her hands and putting them in the shape of a heart as she smiles happy. It will come up. It'll take what you told it and come up with a whole next level of what this should look like, and it'll give you multiple different images to choose from. In Chat GPT Enterprise, these are completely unlimited, and you have full ownership of these images. You are welcome to use these on your socials, on your Twitters, anywhere you want on your website. We give you full license to these, which is pretty powerful. I'll just show you some examples of Dolly 3 and what this looks like. These are some pretty powerful images that are generated on just the first try. Of course, the more you get it specific on what you want to see, a tomato man with tomato veins all over his body, it's going to give you even more specific images. You can always tell it to tweak it. You can give it different instruction, but it's really powerful. Then finally, starting to wrap up here is Chat GPT Enterprise can now access the world in real time. Previously, we had the limitation that it was only trained up until data for 2021, so it didn't know about anything that was current events. That is no longer the case. You can now ask it any question that you want, and it'll go through, it'll pulse the internet for it, and it'll actually show you the sources so you can double check that they are credible. Here's just a small sampling of what's possible with Chat GPT Enterprise. It looks like a lot, but this is truly maybe 5% of the use cases that we're already starting to uncover. As you can really see, this is becoming everyone's personal AI assistant chief of staff. I'll wrap it up here by saying we are just getting started. There's so much more to come, and I really appreciate you all spending time with me today. Looking forward to what's next. Maggie, I'm sure you get this all the time, but after spending 20 or 30 minutes with you, you have to be blown away with what is possible. The imagination just runs wild. That's going to be a great part of your job. The best. Well, we have many different sized companies here today. Kind of the sweet spot today is about 10 to 100 million, but several enterprises also. And you mentioned the Boston Consulting Group research, and they presented earlier, and they said the majority of enterprise Fortune 500 companies have done some level of proof of concepts, pilots, et cetera. They're really talking about really using some of the foundational models, right? But only 10% have actually went into a production commercial utilization. It seems like Chat GPT Enterprise really gives you Chat GPT out of a box, so people can get started right away without having to do a lot of pre-planning. Is that correct? That's correct. For any of you who have used it, or for those of you who haven't, it will take you about 15, maybe 30 seconds, generously, to get started. Literally, you just create an account and you get going. You can download it on your phone. It is downloaded within seconds, or as fast as your Wi-Fi connection is, and you can just start using it. So there's really no type of overhaul, no type of implementation to start using the services. With most software, there is a rollout process, and the training is so important, because I view Chat GPT almost like a blank canvas. I could do anything with it. But if I'm a sales development rep, or I'm a content marketing person, I almost want to have some instruction of start A, B, C. Is that what the templates does? And do you recommend companies kind of put a playbook together? Yeah. And actually, more than anything, we help you with that playbook. We have a team of world-class customer success managers who have deep experience rolling out enterprise software. So our team comes in and works really closely with you. We've got dedicated, prioritized support. We set up Slack channels. We've got the playbooks. We do the trainings with your power users. We love to employ a train-the-trainer type model, because we obviously can't be doing trainings every single day. And we will continue to scale this out. We do realize that it's very much a blank canvas, and if you don't know how to use it, you're going to sit there and be like, I don't know what to do here. The templates is another great start to that. If you don't know what you're doing, and you see all of these templates, sales-related templates, summarizing my call notes, writing a follow-up email, summarizing my recruiting notes, and sending next steps to the candidate, that will help to get your brain going, and you can jump in right away and start using it. Now, when I hear the word Chachi Boutique Enterprise, I automatically think, oh, this is for large companies. Is there a profile of a company, an ICP, that you're looking at, or is it any size? We are rolling out, we're public about this, but we are also rolling out a self-serve product coming up here pretty shortly, and that is going to be really meant for smaller companies, maybe smaller teams within a big company. It'll have a lot of the similar functionality, of course, still continue to be quite differentiated, but that is really going to be meant for the smaller SMBs. And then, I would really say our Enterprise product is really meant for companies that are about above 150 users, where they really start to need things like single sign-on, they really start to care a lot more, not that smaller ones don't, but a lot more about security and white-glove service. We've been seeing most of our companies are buying a seat or a license for every single one of their employees, so I would really say that 150, 200-person company and up is a great fit for Enterprise. I know it's really hard to be asked these questions that we didn't script or prepare for, but since about 50% of our audience do have finance responsibility, are there any one or two examples of how finance organizations are leveraging ChatGPT Enterprise that you could share? Yeah, absolutely. Great question. So, for example, our internal team is now using ChatGPT Enterprise for their end-of-month close, so we're able to take all of our data from Stripe, all of our data from Salesforce, all of our RevRack information, everything that's been booked, and using advanced data analysis, it'll take all of this and it'll summarize and close out our books pretty much for us. Something that used to take us, what I've been told, two to three weeks is now done in less than a day. Now, Wayne, I hadn't thought about this. One of the biggest challenges for SaaS executives, they have all this data from many different systems, Salesforce, Outreach, their financial management platform. You're telling me that there's actual use cases where people will just send all that structured data without knowing the structure and will actually allow you to visualize it and do a lot of the analytics? That's exactly right. And it'll actually ask you questions. If there's things that flag out incorrectly, it'll say to you, hey, this column seems to not be matched. What does this mean? And it'll engage in a back and forth dialogue with you, so that way it strives to get the answer correct. And then, again, you can always see all the work it did behind the scenes, but going back to that FinServ company that I talked about, I mean, they are sending in so much data to the point where they were saying Excel couldn't even handle the data they were sending in and ChatGPT Enterprise couldn't. The final question without leading the witness, seeing all the enterprise deployments that you have, even though it's only been out for a couple months, are there any kind of common cautions or preparation advice when a company's first considering doing this that they should really think about before they start the actual, let's go? I would explain to your companies, I mean, we'll partner with you on this, but the how and the why. For any software tool out there, deployments usually don't go that great when you just turn something on and you say, it's available on Okta, go get it. Employees want to be enabled and trained on things. They want to know what are a couple of use cases. So there is, I can near guarantee every company at this point now has at least a handful of people who are very excited about AI. And so I recommend putting together this tiger team of folks internally where there are different representations across different functions within the company, sitting down with them, understanding what are the common problems that they're running into and what can we leverage ChatGPT for to make their lives so much better and more productive. I'm going to let you answer. There was one more question. One of the anecdotal objections I hear all the time is, I don't want them using my data. I want my data, I don't want my data to contribute to the foundational model. So with Enterprise ChatGPT, you have specific security. That's not going to happen, right? And how do you get the CEO and CFOs of the audience comfortable with that? Yeah, great, great question. Not only have we created a security white paper on this, we've also engaged third-party auditors to prove that we are not training on the data. But just want to be very clear for ChatGPT Enterprise and for our APIs, we do not and we will not ever train or look at the data. Essentially, what we are doing with, for example, the APIs is we are running classifiers to the backend looking for things that are against our terms of service. People do some weird stuff out there, and we need to make sure that we are capturing that, but we are not looking at it. We are not training on it. We are not even touching this data. And then we are deleting it after a period of time. Last thing, I'm sure one of the biggest challenges has got to be, let's say we have 1,000 people watching this live right now. They're like, I got to find out more about this. How do they reach out to you, Maggie? How do they not become the 10,100th lead to know that they can actually engage you? How do they do that? Yeah, great question. So right now, the best way to do it is via our Contact Us form on our website. I manage the team that responds to all of those. We are candidly still working through the backlog, but we are going as fast as we can. We strive to get back with you, back to you within a couple of days at the most. But if we don't, please be patient that we are working our way through the backlog. But we would love to chat with you all. Happy to move as quickly or slowly, ideally not slowly, because people really don't want to sleep on using AI today, but happy to move as quickly as y'all would like. If people want to follow you, because often people will do a session like this and say, well, I need to follow Maggie. She probably knows what she's talking about. Is LinkedIn the best place to follow you? Yep, LinkedIn is the best place. I have a Twitter account. I'm not too active on it. So LinkedIn is absolutely the best place. Okay. Well, thank you very much, Maggie Hott, for speaking here at SaaS Metrics 23. I cannot tell you how much we appreciate it. And we can't wait to see all the future coming out with OpenAI. Same. Thank you all for joining me today. It was great to have you all here. Okay, everyone. Our next session will start in just a few minutes.