Modules
Session 7: How CROs Should Engage with AI w/ Andy Pitre
Transcript
Yeah. As, was said in the intro, my name's Andy. I work for HubSpot.
One question I have, I know we have a lot of great HubSpot partners in the crowd. How many HubSpot customers do we have in the crowd? Okay. Nice.
That's about, like, that's about, like, half the room. That feels really good. And and over here too? Alright.
I love that. So, welcome to Southbound. Thank you all for coming. Thank you guys for having me.
I'm here to talk about AI, and I wanna start, by talking about AI with this statement here, which was said by Pat Grady, who's a partner at Sequoia.
He's much smarter than me, has a lot better industry insights than me. But Pat believes that we are standing at the precipice of the single greatest value creation opportunity that mankind has ever known. So that's a pretty bold statement to make. And he said it in an event called AI Ascent, which Sequoia held earlier this year.
This is just a straight screenshot that I took from the video. I have the link to the video here. It's actually a great, about thirty minute video if you all wanna watch it. But the reason that he says that is in these three bubble charts that we have up here, this bubble chart here is the total addressable market of SaaS back in about twenty billion of that had been recognized.
Right? Fast forward to twenty twenty two, twenty twenty three, and you have about, total addressable market of about six hundred and fifty billion, so the market got way bigger for SaaS. And about four hundred billion of that had been addressed with all the products that were being sold in the SaaS space. But really what you're doing with SaaS is you're taking like on premise software.
Right? Like on premise software and you're, you know, basically converting that into cloud software. So you're replacing software with software. And what's really interesting about AI, especially generative AI, is it not only has the ability to replace software, but it also has the ability to replace, like, the work that we all do as people.
It has the ability to replace services.
And when you think about the TAM, not just of software, but of the actual people and the services and the work that we all do, that's trillions of dollars worth of TAM. And so that's why why that matters. And that's why I think people are so excited about AI right now, because it is this huge value creation opportunity for businesses. And we you know, the people in this room and a lot of, obviously, other smart people outside of this room are gonna be the people who get to figure out how to create that value for our companies and for the companies that we work with.
So that's one of the things I think, makes AI very, very interesting. The other thing I think about a lot when it comes to AI is Amara's law, which says that people tend to overestimate the short term impact of new technologies and underestimate their long term effects. And I think, like, a great example of this is, like, in the dot com, you know, the early dot com, you know, bubble that burst. Right?
People got very excited about the idea that the Internet was gonna change how people buy, change, you know, behaviors, change the way businesses sell. They made lots of investments in the late nineties. They expected really immediate results, in terms of, like, revenue from all the investments that they made. And then they basically said, oh, shit.
These investments aren't paying off fast enough, and you got the dot com bubble. And I'm not necessarily saying that there's gonna be an AI bubble because I think that people are smarter now because a lot of the same people learn from the dot com bubble. Although, who knows? We don't we don't can't predict the future.
But I think that right now, we are definitely in the phase of overestimating the short term impact of AI, and that's why you see things like people making fun of Google for telling you to put glue on pizza and, you know, and make fun of ChatGPT for like bad answers that ChatGPT gives. We are in the very, very early stages. But I think the other other point I wanna highlight here is I don't think that we've even for for as excited as everybody is about AI, I think we're probably still underestimating how big of an impact it's gonna make in our lives on, like, a ten year scale or on a twenty year scale.
And so here's here's just another example of that. This is in nineteen ninety five, the most valuable Fortune five hundred companies in nineteen ninety five. You have GE, Exxon, AT and T, Coca Cola, Great Atlanta Company, Walmart. Right?
Those are those are the most valuable companies in nineteen ninety five. In two in twenty twenty three, it's Apple, Microsoft, Google, Amazon, Nvidia. And I think now Nvidia is basically competing for like the top, you know, the top spot. That might be like the first they're number one now, someone in the audience is telling me.
So they're competing for the number one spot and they're obviously doing that on top of AI. And so I don't think it's crazy to think that in twenty years, you know, we basically moved from, you know, all of the top companies being software companies to all of the top companies twenty years from now becoming AI companies. Right? And that's kind of what we're at the beginning of.
And we are very, very early, right, in this entire process, and there's good things about being early, and there's bad things about being early. The bad things about being early is, like, you can't predict the future. Right? Predict like, like, Yogi Berra expression is right.
Predict predictions are hard, especially when they're about the future. Right? And so we don't know exactly what's gonna happen. It's impossible to say what's gonna happen.
We're gonna make a lot of mistakes collectively. All sorts of companies are gonna get created, build things. Those things are gonna fizzle. You know, some of them are gonna work.
Some of them are not gonna work. Probably more of them are not gonna work than are going to work. But the advantage of being early is that, you know, we have that first mover advantage, right. It's the people in this room who are going to be figuring out the technologies and the use cases of AI and some of them are not going to work, but some of them are going to work, and the ones that work are going to be really transformative.
And so I think that's a good thing. And I think about an analogy, to, you know, where we are now to like the early days of HubSpot. So this is a, I joined HubSpot in two thousand and nine. This is a screenshot of what the HubSpot product looked like in two thousand nine.
Were any of you HubSpot customers like way back, like ten, ten years ago? We got, okay, we got a couple people. You guys remember Keyword Grader? This was like the original killer feature of HubSpot.
When I joined HubSpot, we had a few different features. Most of them were not very good, but Keyword Grader was really, really good. And the thing that Keyword Grader did, even though it didn't have a great UI, was it basically helped companies determine long tail keywords that they could rank for on Google. So it was an SEO, tool.
And what's interesting is HubSpot started in two thousand and six. I joined HubSpot in two thousand and nine. That's basically about a decade after, you know, Google and search engines became, you know, a thing and, and slowly took over as the way that people use the Internet. And so at the time when I joined HubSpot, I actually thought that we were like late to the game.
Right? Like I joined HubSpot and I loved HubSpot and I was a big HubSpot fan before I joined HubSpot. I read all the marketing content. I read whatever.
But I kinda thought like the Internet was done. The Internet's, you know, it's like it's been around for, you know, twenty years. Search engines have been around for ten years. Like how much more innovation could could there be?
But what was interesting is even ten years after, you know, Google came out, there were still millions of small businesses that didn't have web original thesis of Hub they, you know, I I had so many customers back in the day who literally were like, I used to just advertise in the Yellow Pages and nobody uses the Yellow Pages anymore, and I have no idea how to show up in Google. That was like the the class Yellow Pages anymore, and I have no idea how to show up in Google. That was like the the classic HubSpot customer back in two thousand nine. And so even though it was ten years, you know, after search engines, even ten years later it was still early.
So that's how, you know, now we're two years after ChattGPT, barely. Right? Like November will be the anniversary of like two years after ChatGPT. And so we are like so, so, early.
And the other reason I kinda show this and talk about the early days of HubSpot is because one of the theses of Brian and Dharmesh, the two co founders of HubSpot, was that although businesses were not doing a very good job transitioning from offline to online, there actually was this huge advantage or this huge kind of, ability to level the playing field for small businesses if they did move online. Because, you know, before Google, you know, your advertising budget was was basically the thing that determined, like, how how wide you could reach. Right? And so if you were a big company, you had a big advertising budget.
If you were a small company, you had a small or nonexistent advertising budget. And so you were at this huge disadvantage. But when search engines came along, suddenly it was like, oh wait, as a small business, if I create a website, if I create a blog, if I create content, none of that is incredibly expensive to do. I can show up just like any other big company can show up on Google.
It was this, it was this, moment where you could level the playing field for small businesses. And that was what really inspired, Brian and Dharmesh to start HubSpot and get into the inbound marketing game. And I think the same thing is true for AI. Right?
Because I think the dynamic that you have today is big businesses can hire lots of people. They can have big marketing teams. They can have big sales teams. They can have big services teams.
And small companies are always at somewhat of a disadvantage because they can't bring in as many people. And what the power of generative AI has the capability of doing is basically making it so that you don't need a huge team anymore. You don't need that big marketing team, that big sales team, that big service team, because people are going to be so much more efficient within the organization as they use AI powered tools. And and just to like make this case, this is, this is like not taking into account AI, but this is a really interesting chart.
This is the number of employees that were required in a business to generate a million dollars of revenue for that that business. So if you go back to like nineteen ninety, it took about seven employees to generate a million dollars worth of business. And if you come down to like twenty twenty, you know, twenty twenty two, twenty twenty three, wherever this chart ends, it's now two employees to generate a million dollars worth of revenue for a business, and that's before AI. So that's just the, efficiency that was gained by adopting SaaS software to make people more efficient at doing their jobs.
And I think what becomes interesting about this chart is I think with AI, there's gonna be, like, a whole other curve under this, and you're gonna probably have to change the axis of a chart like this to basically say, like, how many employees is it gonna take to make, you know, ten million dollars or a hundred million dollars. And I I heard kinda through the grapevine that there were some discussions going on at some AI companies around, like, would we ever see a one person billion dollar business, like, in the future? That's I I don't know if that's necessarily true, but that's, like, an interesting thing thing to think about when you can start a business and you can automate so much of the work that's required to drive the business, that you could actually have these incredibly small teams, you know, building building huge businesses.
And I think the thing that's gonna that we're gonna see in general is we're going to see, like, we're gonna see more small businesses. Like, that's just kind of the bottom line for me. We're gonna see more small businesses being created, and we're gonna see less big companies. And I think we're already seeing the beginning of big companies realizing that they don't have to hire as many employees.
Some companies are doing layoffs of employees because they realize that they don't need all of those people. What's ultimately gonna happen is, you know, it's obviously not great to see layoffs, but those smart people who are leaving big companies, at the same time that that's happening, you also see the barrier to entry for starting a business going way down. Right? Because again, you don't need as many developers, as many sales people, as many marketers, as many service companies.
And so I think and I hope, as someone who works at HubSpot and loves small and midsize businesses, that we're gonna see a real rev, real like renaissance in small businesses being created.
The other thing we think about at HubSpot is obviously about front office teams. So the marketers, the sales people, the service people, the rev ops people. Right? Those are kind of the the four core personas that we serve at HubSpot.
And the work that all of those people do is going to absolutely, radically change. Right? So what we're already seeing in marketing, and this is in the HubSpot product. We have a bunch of Gen AI features, especially, built around creating content inside of HubSpot.
And so for the people who are creating content, emails, blog posts, social posts, you know, any sort of that content that you would create, about seventy percent of HubSpot customers are now using generative AI in some way, shape, or form, inside the HubSpot tool in order to create those posts. So we're already seeing this huge transition of marketers being more efficient from AI. We also, are gonna see sales reps becoming more efficient with AI because a lot of the work that sales reps do is around researching the companies that they're gonna reach out to, finding new companies to reach out to, figuring out what the outreach, you know, emails and sequences are gonna be to reach out to those companies.
And a lot of that work today is, like, really, really time consuming. But in the future, that's work that sales reps are just gonna be able to rely on AI to basically be their BDR, to find them those prospects, to reach out to those prospects, and then they can devote more of their time to the things that they're really good at, which is actually, like, building relationships and closing deals. And so I think we're gonna we're at the beginning of seeing that. And for customer support, the thing that we're already seeing at HubSpot is, you know, we've implemented, you know, our own tools for generative AI to answer support tickets for our customers.
Service person, which is really great. And it's a big win for having to get rotated to a service person, which is really great. And it's a big win for customers because they don't have to wait in a queue to get an answer from a from a service person. And it's a great thing for us because we don't have to scale our services team, as much as we would otherwise. So it's a great thing for our, you know, overall, operating profit.
I think the way that we're gonna see AI manifest, and I wanna I I'm a product person, so I brought some demos. I'm gonna walk you through some HubSpot demos today.
But but I think the way that we're gonna see some of these things manifest is in an AI world, especially if you're doing anything customer facing with AI, the value, like, the the accuracy and the amount of customer data that you have have is gonna be incredibly important. Right? So today, you can go to ChatGPT, and you can say to ChatGPT, you know, write me a sales outreach email. And it'll do a pretty damn, you know, kind of scary good job, like, writing a sales outreach email.
But the more you can feed the model in terms of, like, who you're trying to outreach to. So if I say, build me a sales outreach email for somebody who's a Boston Celtics fan, for somebody who's been on my pricing page, for somebody who's been on this industry, or somebody who works inside this industry, for somebody who has this title, for somebody who's had these previous interactions with me. Right? Like, the content that you can create with AI gets better and better and better.
So the ability to have that customer data altogether all in one place is gonna be huge, and that's something we think a lot about at HubSpot. That's what, you know, we really want HubSpot obviously to be, you know, the central place for all of your customer data. And there's obviously a lot of other smart, great companies that are trying to do the same thing, but hopefully, we'll find a way to do it better.
The other thing that we're gonna see is in software, we're gonna see more and more Copilots.
And the name still might change, but it seems like Copilot is becoming sort of the industry standard term for, like, an AI, you know, helper buddy that, like, lives inside of software. You see that with, Microsoft Copilot for writing code. We have a Copilot inside of HubSpot called ChatSpot that we're building out to help you do your work. And those CoPilots are gonna help you work more efficiently.
So that might be doing that sales, you know, might go to your CoPilot and say, hey. You know, who are the prospects I should reach out today? What are the next steps I should take for those prospects? Can you summarize my activities that I've had between between me and those prospects today?
You're gonna see co pilots doing that type of work. And then finally, I think that beyond co pilots, we're gonna eventually evolve to a world where you have full on AI agents that are basically doing the work for you. Right. So going to an AI agent, basically giving it a task to say, say, build me this marketing campaign to drive this outcome.
Right? Find the prospects for me and do the initial outreach for those prospects for me. Right? Take these support tickets and be able to automatically answer those support tickets.
You're starting to see the early versions of that today, but that's really where we get into the world of AI starting to replace human work. And the way Dharmesh talks about it, which is obviously scary because we're obviously the humans whose work is gonna potentially get replaced by AI. The way Dharmesh, who again is the cofounder and CTO of HubSpot, you know, know, he talks about this idea that, like, AI probably will replace some of our jobs, but it will hopefully give us better jobs at the end of the day, which is something I truly believe. So the work that we do, is gonna be way more focused on the strategy, way more and way less focused on, like, the mundane day to day tasks that we have to do because AI is gonna be good at those things.
Okay. So I wanna show some demos that kinda hit home some of these ideas. The first one is a demo of, chat spot inside of HubSpot, which is our Copilot running inside of HubSpot, and how that can help with things like sales outreach, inside the product. And again, I'm I work at HubSpot, so I'm gonna show HubSpot software, but there's lots of other great tools. I don't wanna be overly promotional. There's other great tools that do things like this, but this is just one example.
HubSpot, you'll be able to access your Copilot anywhere you work, from not only prompts, but we're also creating a handful of in app hooks that allow the user to click one button and get results from their Copilot. So from a company record or company index page rather, I can get a summary of any company. I don't need to visit the record. I don't need to open up a new tab. I get a quick summary with all of the detail on that record directly in the Copilot.
Now now that the summary is provided, I can ask the Copilot for recommendations.
So what action items would you recommend?
Now this is a really powerful combination of the data that is stored in the HubSpot CRM and the power and knowledge that an alum has. So we can create action items that are tailored to those engagements, directly forming, right, in Copilot. So I've got a few action items here.
So I'll pause I'll pause on that one for a second. But the idea there is with all that data inside your CRM, one of the things we see all the time if you shadow salespeople and we've how many salespeople do we have here in the audience? Probably a lot. Right?
How much time do you spend doing that research? Right? You go into the CRM, you look up the company, you leave the CRM, you go to their website, you look for news, you look for updates, you look for, like, anything that you can get from that company to basically be more relevant in the way that you're outreaching to them. Right?
That takes a lot of time. Right? And if you're a good sales rep, you can do that pretty quickly, but it still might take, you know, ten minutes per prospect, you know, twenty minutes, half an hour per prospect depending on how much time you put in. That's work that can all be automated with AI now.
So you can get that summary. And within HubSpot, we're looking at not only the data inside the database, but we're also starting to look at data outside on like the open Internet about that company, pull that in, give you the quick summary as a salesperson. So it's almost like having an assistant who's gonna do that work for you. That's the way that we think about Copilot.
And then you can just ask Copilot what the next steps are gonna be. So you can basically say like, okay, well, what would you recommend I do here? Now you still need a human to evaluate some of the things that are, you know, coming from AI, but it's definitely making people's lives a lot faster. Now on the marketing side, this is a demo of a tool we have in HubSpot called Content Remix, which launched a few months ago with our new, Content Hub product.
And so this is a really cool tool that helps marketers basically be, build and then repurpose content faster.
I'll start by just doing a brand new remix and let's go in here. I'll select one of my blog posts.
In this case, let's do a blog post fittingly about repurposing content, and then I'll move my face in Loom, add content, and it's got to load in that blog post. Now what we can do is go in here, and one of the first pieces of or excuse me, one of the first new features that we added in is support for the email content type. So this is geared toward marketing email copy where we can go in and add that. Let's hit next. Let's say that the objective here we're going to encourage recipients to read the original blog post.
This is to our marketing email subscribers and the writing style.
Let's make this, witty and friendly.
And we'll hit generate there. And as soon as we do that within just a couple of seconds, the content remix tool is going to be able to generate us a really solid marketing email.
Now with the way that this generation works currently, what it's going to give us is the body copy for a marketing email. And it's formatted and marked down so that we can just copy this and put this into our marketing email tools. Soon what we'll be able to do is use the save and edit in app functionality. Now speaking of that save and edit in app functionality, we have created that support for the blog post. So let's do this, and let's say focus specifically on the use case of repurposing industry research.
And we'll make it short so it'll be just a little bit quicker.
One thing you may have noticed is that the longer or more complicated the content that you generate in Remix is, the longer it's going to take. That's just a byproduct of the models that are at play here. One of the things that's always on our radar is how do we make this faster, how do we make this, you know, even more of a sleek experience.
The one thing we never want to sacrifice on is of course the quality of the content that we're generating.
So in about forty seconds we So I'll fast forward on that.
But basically, the blog post gets generated and the idea here of content remix is you can take one piece of content that you've created. You You can even use AI to help generate that original piece of content, and then you can repurpose it across email, create more blog posts, create social posts, create ads, do all sorts of stuff that a marketer would typically do to run a marketing campaign. And again, these are things that might have taken a full day for a marketer to do or possibly like multiple days for a marketer to do. You could now do do these things in seconds or minutes, right, inside of HubSpot.
So this this is the way that marketing is gonna continue to evolve. Now you're still gonna need really smart marketers who are gonna be able to direct tools like this to understand what they're gonna do, but you're gonna be able to have, you know, smaller marketing teams or agencies that are able to do this type of work for you without having to hire big marketing teams yourself. So that's the direction that we're going. And then this demo here is showing our, support agent and how you train the support agent on content to be able to answer questions.
Hello. Today, I'm gonna walk you through the onboarding flow for the AI chatbot.
On the first screen, users will select from HubSpot knowledge bases or import external URLs to train their bot bot.
Once the bot is trained, it will use the information to respond to questions it receives based on the content from these sources.
I'm going to add an external URL here from cell NYC. Basically grab any content from your website, from your knowledge base Once we add the source here, it will be added to the queue for our bot to begin training on.
Next, users will be able to select either a help desk or inbox chat flow to assign the AI chatbot to. I only have one chat flow here for help desk, so I'm going to select that one.
Last, users will be able to set a custom welcome message. This is the first message users will see when they open the chat flow. They'll also be able to select users or teams to reassign Once I hit assign to live chat, it will bring me to the previously selected chat flow settings, and we should see the input welcome message as well as the assignment to our AI chatbot with the fallback assignment to the selected agent.
Now let's go check if our source has been synced.
It has not yet. So I'll come back a little later.
Now that our source has been synced, let's check out the chat flow on our test landing page.
If I open the chat flow, I should be able to ask a question, and the bot will be assigned to respond.
Even though the bot has answered the question as we'd expect, I'm going to say it didn't answer my question to demo reassignment.
So I'll skip ahead on this one a little bit too.
But basically to me and the way that we think about things at HubSpot is this is really the future of support.
Right? So we're using this tool ourselves to basically deflect about twenty percent of our cases. Right? The things that could be solved by knowledge based articles.
You still obviously have more in-depth cases that are still gonna need to get routed to internal support people, but you can handle a lot of those, like, really, really easy use cases product, and this is also, free right now to create one agent with the service hub product, which is, you know, pretty crazy. These these things are gonna become you know, we're not the only company that's building bots like this. Right? Other support companies are building these, but this is the future of where support is going.
And then I have one more demo, and I think we're gonna have some time for Q and A. This is a non HubSpot demo, so the other ones have been HubSpot demos. This is a demo of a tool called HeyGen. Has anyone here played around with HeyGen?
So HeyGen basically takes a short video that you record of yourself and turns it into an avatar. So what you're gonna see here looks like me, but I didn't actually record this video. This is the HeyGen avatar of me, talking to everybody here in the room.
Hello, southbound.
This is not the real Andy Peter. This is just my avatar. HubSpot is working on an integration with HeyGen where you can create video of yourself based on the content you create and your customer data. Isn't that cool?
So you can see it's, like, not really perfect. Perfect. I pronounced my last name incorrectly. Right?
Like, it's it's it's not, you know, it's not quite there. But it's, like, scary that this is technology that exist today. And if you think about combining that with the ability today. And if you think about combining that with the ability to understand your customer and create content, you know, basically directed personally to your customer, this this is, you know, again, the future that we're like moving towards with with AI.
So, I wanna get to the q and a and I wanna give us a little bit more time for it, but my my CTA is just at the end of this for everybody. My calls to action are one, if you have not already, and I imagine most of you probably have, go get your hands dirty playing around with different AI tools. This is a bit of an eye chart, but, but but there's a bunch of really cool tools out there that you can start playing with. Hey Jen is one of them.
One tool that I think is obviously, you know, the eight hundred pound gorilla in the room is ChatGPT, OpenAI. They have a really, really good playground. I don't know if anybody here has like signed up for a developer account with ChatGPT, but you can go into the playground, you can create different instructions that you can give to, like an AI agent or an AI co pilot, and then you can give it different prompts. I use this myself to test out, you know, different half baked ideas that might be interesting to, like, build into HubSpot, because, we use OpenAI as technology underneath HubSpot.
So this is a great thing to play around with if you really wanna start getting your hands dirty and learning AI. And my my sort of, big takeaways from all of this today is I think we're at a revolutionary moment in software and in how people work and how people buy.
We today, the people who are in this room, we are the early innovators, and there's advantages and disadvantages to that, but I think the advantages outweigh the disadvantages.
Really good customer data is gonna be incredibly powerful for any AI, use case that's directed towards customers and obviously, all of us here are thinking about how we build things to serve our customers and our customers' customers.
I think everybody should start experimenting with AI. Like we all are sort of at this at this early point, we have the ability to become AI natives, right, if we if we get our hands dirty today and we start working with AI. I think that SMBs, and this is this is good for me and I think good for most of the people in this room, they have the biggest opportunity with AI to level the playing field. I think we're gonna see, you know, tons of SMBs being created, tons of SMBs growing faster than anybody would have thought they could have grown, in the past.
And I think that, you know, for all the doom and gloom around AI, I actually think that the future is bright. I think that AI is gonna be a technology that really serves all of us well as people. I don't think it's gonna replace our jobs or if it does replace our jobs, we're going to find more interesting work to do. So I think the future is very bright in an AI powered world.
And that was my presentation, but now I think we have how many minutes for Q and A? Ten minutes for Q and A. So thank you all very much.