Summary

Jonathan Corbin, the founder and CEO of Maven AGI, is backed by some of the best investors in the industry, like Sequoia, M13, and Lux Capital. Jonathan shares his journey from managing 1,000 people at HubSpot to starting his own AI startup, where he got together with Sami Shalabi (Founder of Google News) and Eugene Mann (Product Lead at Stripe). He discusses the evolution of customer support and the role of generative AI in improving customer experiences. Corbin emphasizes the importance of ingesting and analyzing data from various sources to provide personalized interactions and take complex actions. He envisions a future where customer support is streamlined, siloed systems are eliminated, and AI agents have contextual knowledge of customers. Corbin also advises aspiring entrepreneurs to focus on solving the actual problem with a first-principles mindset and leverage their specialized knowledge and skills.

Takeaways

• Generative AI has the potential to revolutionize customer support by providing personalized interactions and taking complex actions.

• Ingesting and analyzing data from various sources is crucial for creating a differentiated customer experience.

• The future of customer support involves breaking down silos, eliminating ticketing systems, and providing a single touchpoint for customers.

• Entrepreneurs should focus on solving the actual problem and leverage their specialized knowledge and skills.

• The future of generative AI is exciting and has the potential to transform various industries.

Alp Uguray (00:01.122)

Hi everyone, welcome to Masters of Automation. Today I have the pleasure of hosting Jonathan. Jonathan, welcome.

Jonathan Corbin (00:08.94)

Yeah, great to be here. Thanks for having me.

Alp Uguray (00:11.216)

Thanks for joining. Who would have known that I would run into you at the MIT event after playing basketball? The roles are colliding and we are both in the Gen AI space. To kick things off, I'd love to hear about how your story was. You were managing about 1 ,000 people at HubSpot. How did you get there?

Jonathan Corbin (00:18.444)

Ha ha. World's colliding.

Alp Uguray (00:41.6)

what may give you the idea and courage to take the leap of faith for Maven AGI.

Jonathan Corbin (00:49.654)

Yeah, I think the story starts far earlier and I know we have a hard time stop so I won't tell the full thing, but maybe I'll give you highlights. I actually started off my career working for a company called Sybase and it was known for creating the SQL database. So I was actually writing code. Now, if you were to ask our CTO today, Sammy at Maven, if Jonathan could write code, he would be like, I wouldn't let him near any code anywhere ever.

Alp Uguray (01:16.56)

you

Jonathan Corbin (01:17.678)

But that was at beginning of my career. So I spent some time there, and then I worked for a company called Unica that was very early on in the marketing automation space. They were one of the first vendors to do that. You mentioned like an MIT event. Funny enough, the founders were one of the MIT Blackjack teams. So I spent some time there. I spent some time at WPP, which is one of the largest ad conglomerates in the world, running a global consulting organization, working with companies like Coca -Cola on their 10 -year digital strategy.

actually helping create that. What do you want to accomplish? How are you going to measure it? What are the systems you're going to use? What are the tools that you're going to go into that? From there, I ended up joining a company called Omniture that I was working very closely with as part of the tool sets that we're recommending to customers who are deploying. And so Omniture ended up getting acquired by Adobe and I spent five years there. Started off as an account manager, working with some of largest accounts that they had, and then ended up owning the eastern half of the US, Latin America, and Canada.

I helped start the financial service and insurance vertical. And then after five years there, seeing the impact that the software was having on the business we're working with, I decided, you know what, I'm going to go start my own company. And maybe that's how all ill -fated journeys start off. not sure. But I started a company called Veer and what we understood from the product that we created, which was an app to help drivers find the closest seat to the park was that we didn't understand where people were going.

what their destination intent was and we could engage with them around that. I ended up selling the technology to that and did the thing that most entrepreneurs do when they have a sale. They do a little bit of travel, see the world. And then I came back and I joined a company called Marketo. I spent about a year at Marketo. Then we ended up getting acquired by Vista Equity Partners. And then of course they ended up selling that to Adobe. I ended up joining a company called Sprinkler. It was a startup based out of New York that was about 130 million revenue then.

ended up staying with them to about 300 million revenue. Of course, they IPO'd a few years ago. I say it's it for them. And then I started HubSpot. so prior to starting Maven, I spent the last four years from 2019 to 2023 there. And we went through this like massive period of growth there. It was like pretty phenomenal. It was like five companies in the world that were growing at that rate, which is crazy.

Jonathan Corbin (03:40.75)

We were one of the five fastest growing B2B SaaS companies ever after a billion dollars in revenue. And when I started off there, my team was about 238 people. And as you mentioned earlier, it was about a thousand by the time that I left. Over that time period, we went from 70 ,000 customers to 220 ,000 customers. And just to give you context around the number of customers, Salesforce has 150 ,000 customers today. We had three and a half million premium customers. So it was pretty amazing. It was just a great period of time to be there.

where we had these cores, we were growing 83 % core over core. And it wasn't quite to the fact where we were like, know, grabbing people off the street, handing them laptops and phones and being like, take calls for HubSpot customers. It was like close to that though. And it was interesting because when I started 2019, of course, like we all like lived through the COVID times. And so like started things were going well, COVID all of sudden, like we thought the entire world was going to collapse along with everyone else. We went from like, you know,

having plants that we watered every morning in the office to like not see in the office for two years. And so we went from everything going down to all of a sudden like this incredible growth spurt. And so it was just a fun time to be there and to live through that. There was some great leadership that was in place while I was there. I started off working with the founders, Brian Halligan, Darmesh, of course, Yamini came in and she's the current CEO. I to work with some great folks.

market teams like G2 Mitani. I was based out of here in Boston as well. And yeah, really fun time. And then of course I started Maven last year. And so Maven is a Gen .AI startup in the customer experience space. And we're launched publicly in June of this year. We're helping a whole bunch of customers like TripAdvisor, OhamSpot and many others to deploy Gen .AI for their customers and help to improve that. And we've created this Asiatic framework where we're answering as much as like...

93 % of their customer increase with no humans involved. So that's the short version now. I'm glad you asked for that at beginning, because we'd still be talking about me.

Alp Uguray (05:47.44)

Yes. That's a fantastic story because like you've seen the high growth hotspot and then you've seen also the customer support segment there and then that transition to become even AGI and then like addressing that problem. It is really interesting though because like I've been hearing customer support is one of the industries that is waiting to get disrupted.

And it's also one of the biggest things that from a customer standpoint people hate the most.

Jonathan Corbin (06:24.972)

Yeah.

mean, who hasn't had a bad customer experience, right? I think every time I fly, have a bad customer experience and you're like, wow, I wish that I could just like get these people to use Maven today. It would be such a better experience. I think what's interesting Alp is that when you take a first principles approach to how do you solve customer support, right? How do you solve like shitty customer experiences? You start to realize, well, what do customers actually want when they're going to their support reps?

Alp Uguray (06:37.722)

Yeah.

Jonathan Corbin (06:55.072)

And so when they were coming to, you know, the support team at HubSpot and they're saying, how do I set up a marketing campaign? Is that really what they wanted? Do they want to know how to set up a marketing campaign? Or was the real reason that they were coming to HubSpot and the reason they purchased the product is because they wanted more leads. I think it's the latter, right?

Like you don't buy a drill because you want to drill holes. You're like, you want to hang something on the wall, right? And so it's the same principle that's involved. Like you take the first principle approach, you understand what people are really trying to do with your products and how do you go about solving that? And I think what has happened for us is GenAI has unlocked the capabilities to be able to do that.

And so the way that we typically, the way that we think about the problem is like the first step is being able to answer people's questions, right? That it removes probably 98 % of the frustration, but it doesn't make it necessarily a better experience. What makes it a better experience is if you have a personalized interaction. If the, you know, kind of the AI agent or if the person or whatever you're interacting with actually has full knowledge of all the interactions you had historically.

if they actually understand why you purchased the product, why you decided to bring that into your organization. And then the third thing is actually being able to take complex actions for you. And so, and you think of the evolution of generative AI over the last couple of years, we've gone from like, hey, this thing is pretty magical to actually, it can create magic for us. And I'm really excited about us being able to do that for lot of folks.

Alp Uguray (08:12.879)

Hmm.

Alp Uguray (08:18.527)

Thank

Alp Uguray (08:27.992)

That's very exciting because it started with connecting to knowledge bases and the guides and then really streamlined access to information from human agents perspective too. For example, all of the customer support orgs measure themselves with average handling time and then the NPS score. But NPS scores suck and average handling time is so high because they go through these like

disparate systems trying to find out information and then in

Jonathan Corbin (09:01.59)

about ticketing systems out, I'm curious to get your take. What do you think ticketing systems do and do you think they exist in the future?

Alp Uguray (09:09.614)

I think ticketing systems will all go away, and should go away. Because what it does is it's resolution is just solving that small problem. So it's like, want to cancel my phone bill. And it's like a very, very, very direct approach. And I have the trigger to that. I have the knowledge access, and then I have the action. Like I know my problem will be solved if my bill is.

like stops, they stop paying that bill. So from that perspective, like for maybe tickets that are low complexity to solve, I think they're going to go away. I mean, that's mostly like you said, like 99 % of tickets are low complexity. Like people just want the resolution. What do you think?

Jonathan Corbin (09:41.357)

Yeah.

Jonathan Corbin (10:03.276)

And I don't think it needs the. Yeah. So I think there's like, I think there's two separate categories here. One is ticketing systems by themselves. If you ever think of like a Zendesk or, know, like Salesforce, know, like those are ticketing systems. And so what happens is like, you and I are pissed off because our United flight just got canceled and we're like, what the hell guys? Like, I need an answer here. So we're like, we're like shooting them texts, you know, through whatever mediums we can find them. They might interact with this with.

And so like what happens is that it goes into like a Zen desk and it like, okay, what's the, you know, what's the right person to be able to answer this? And there's different ways that people will route tickets. You know, typically it's either like language capacity. And so one of the challenges that I had was finding great people who spoke Japanese that have a spot.

And there was lots of people that spoke Japanese. The challenge was finding people who knew HubSpot, they were willing to switch jobs and all that stuff. took us a little while to make that happen. The other side of people that spoke German and Italian and stuff like that. And so if you were to interact with an agent, you were to send them a message in German, you had to find someone to respond in German. The other thing is sometimes there's specialized information that you have to be to answer a question.

So those are kind of like the two reasons why you have taken systems. And so you think of like what's happening right now where you can create these AI agents that can ingest any kind of information. Now like that, you can actually speak multiple languages, right? So it's like, okay, you're starting to break down the barriers and the reason why those systems existed in past.

The other reason you could probably argue is like, what about reporting Jonathan? Like not everyone is using a BI system. Maybe you just like once the reports out of box. First of all, like I think static reports are dead. Like no one's going to use those in the future.

Jonathan Corbin (11:55.234)

The future is actually dynamic reporting where you don't want to look at something and see like a blip on the screen and be like, I wonder what that was. Yes. You want to know what's going on in your business. Right. So you would have a conversation with your technology, just like you would with a human. If you're talking to anyone else, you'd be like, how's my business performing? And they're like, great. But we had this like thing where people were locked out of the system for an hour and a half.

What you actually want is the analysis, not just the data, not just the information. You want to understand what happened, why did it happen, and how do we go about solving it? And so I think the reasons why these systems existed in the past, it's going to be completely gone. So I actually am very excited about that, because I think it's going make for a better experience. I think the second thing is, what are the categories of customers that need to be served through technology?

Alp Uguray (12:19.61)

Mm

Jonathan Corbin (12:45.91)

And I think of them three categories. The first one is people who want, like they want to self help, right? And probably you and I probably fall into that category, right? We're like, the last thing that we want to do is we want to have to like pick up a phone and call somebody or actually like talk to an agent. It's like we'll search for YouTube videos. We'll like do a bunch of stuff, take our shot on like, you know, interacting with chat to get you to see if it has the answer for us. And so it's like, okay, great. Like you want to self help.

Alp Uguray (13:06.606)

Yeah.

Jonathan Corbin (13:14.604)

The second category is like people who are like self help, but they want validation. So like found the answer before I hit this big red button, I want to make sure that it's the right thing to do. Cool. The last category is probably like my mom. And I know she's not going to listen to this. I'm very happy to say that she's like that last category. And she's like, Hey, I have, you know, the last Microsoft phone that exists on the Verizon network. I don't know how to like make it turn on.

Alp Uguray (13:31.492)

you

Jonathan Corbin (13:44.534)

I can't find the red button. I don't know how to put the plugin to it. Like she needs like someone talking her through the whole thing. I think what you find is like you really want to conserve the humans to talk to people like my mom who need that human touch. But the other category is how do we do a better job of assisting them, helping them not just kind of answer the questions, but actually being able to take on those more complex actions.

Alp Uguray (14:09.934)

Yeah, that's very interesting because over time, I think as AI gets smarter and is able to handle more tasks, then it would be able to perhaps facilitate some of the conversations that are deeper or more complicated, even in the relationship building with the customers. It could be maybe lower ticket type to more complicated tasks.

Jonathan Corbin (14:31.842)

Yeah. Yeah. Well, you know what goes into the relationship building, right? And the reason why I'm usually really frustrated during the experience, the support experiences I have is like, there's no context, right? So you might've interacted with them and used their product for many, years. They're like, for every time you interact with them, it's like you've never purchased anything from them before. They have no information about you. They don't know who you are.

Like I think a really good example is like you look at consumer products. I'm sure like we have all purchased probably tens and tens of thousands of dollars of Apple products over our lifetime. But every single time I need help or I need support, it's like, hey, how are you liking your iPhone? I don't know, 14 or whatever it is. And you're like, no, what about the hundred other products that I purchased from you guys? Like you should have me on a list someplace. I should be like VIP, you guys have purchased a bunch of stuff from me.

You should know all that information. I want you to. Same with Nike. We were talking about the gym earlier. I buy a bunch of Nike shoes and gear and I buy basketball shoes every three months and running shoes every three months. So I buy the same things. You know the sports that I engage in. Where I live because you ship stuff to my house. Why can't I have a personalized experience like that where it actually shows you have context around a relationship? For me, that's a much better experience and that's what I want.

Alp Uguray (15:56.816)

It makes it much more personalized and also like saves time as well. So I was calling this healthcare provider the other day and then I wanted to just like ask for an appointment and they connected me to four different agents. And then at the end of it, like at the end of like 45 minutes, they were like, we are not yet no appointment. Do you want to try the urgent care? Like if I have access to the appointment.

If there was like that chat GVD that had access, could have just asked that immediately and solve that problem. There's that big gap between, I think, like the way the organizations are set up. And I'd be curious to hear your take on it because there are the modern companies that use more modern tech stack. Maybe it's easier to plug into their data and then quickly ramp them up. How do you approach to more of a legacy company?

Jonathan Corbin (16:26.082)

What the?

Alp Uguray (16:55.741)

That's like a bit of a crappier systems, not a clean data.

Jonathan Corbin (17:01.666)

Yeah. I think what's interesting and the challenge I think a lot of people are facing and actually like leaning into generative AI is that you can't just like, Hey, let me grab an LLM off the shelf and put it in front of my customers. That's really dangerous. Like it's going to say some crazy stuff to you, right? There's examples of Gemini saying, you're hungry. Maybe you should eat rocks. Like that's not good. That's not something you want to tell your customers. Right. And so.

In order for you to be able to do a good job of creating a JNI experience, you actually need guardrails in place. And for us, one of the things that we found out going into the creation of Maven and deploying it out was enterprise data is really messy. And there's a whole bunch of data silos, no matter which organization you're in. I was most recently at HubSpot. My co -founders were at Stripe and Google. We all had data silos that existed.

And so what we realize is that every single organization has data silos in order for us to be able to do a better job than anyone else does in terms of answering customer questions. We need to ingest more data than anyone else can. So we can ingest any kind of structured or unstructured data. So if your data is living in like a basket under Joe's desk, if it's living on like, you know, a USB key from 25 years ago, or if it's on a dual drive and you have it in a PDF, like

great, we can ingest that. Snowflake and all that other stuff is probably much easier to do, but we can ingest any kind of structured or unstructured data. That's really important. When you're starting to work with folks like TripAdvisor who've been creating content for the last 25 years, you actually need to be able to ingest all the information, but you also need to understand what do you do with all that information? And so I think that's where you start to really dig into.

Alp Uguray (18:50.832)

Thank

Jonathan Corbin (18:54.51)

Creating a differentiated product from the other ones that are out there. I think for a lot of folks who are thinking about GEN .AI and using it in the context of customer experience and customer support, they're like, great, I'm going to take a bunch of data, I'm going to throw it into an LLM, and I'm going to cross my fingers and hope for the best. That's not a good strategy.

Alp Uguray (19:13.722)

And in the generative AI meetup that you invited me, you said something really interesting. Because we're saying customer experience does not just happen. Customer experience, people think about that process, they design it, and then make sure that a customer is able to follow the journey, and then they're happy at the end. That is so true.

Jonathan Corbin (19:37.709)

Yeah.

Alp Uguray (19:41.508)

People think that they can just map a generative AI solution to a data system and then it will magically work. Like designing that CX experience is massive. Like what are you seeing as it is evolution today? Like how do you think the CX experience is going to change and become? And how does like Maven, how do you guys work with organizations, organizations support that?

Jonathan Corbin (20:03.939)

Yeah.

I'm actually really encouraged by a bunch of things that I'm seeing in the market right now, right? There's a lot of the GEN .AI kind of missions being handed down from boards from the C -level team and saying, hey, we need to come up with the GEN .AI strategy.

Hopefully that makes it a little bit more holistic. And I think companies are thinking about it a little bit more in that fashion. Where do the best applications for generative AI capabilities within our organization and how do we get the most value out of them? And so the way you were asking about the way that we work with folks, actually, what we're getting pulled into is a bunch of conversations around exactly what you're talking about. How do we design a differentiated customer experience that's going to help us create a moat for our business?

And so when I go back through kind of all the history of customer experience organizations that I've worked with in the past.

I think people continue to focus on that. And I love that because I, as a consumer, want a better experience. And so people are coming to us and saying, here's how customers are interacting with us today. How do we change this? How do we make this better? How do we make it more concise? How do we bring more value to our customers faster? And so we're actually able to work with them. And some of them have consultants they're working with in terms of helping to draft that strategy, create it, and move

Jonathan Corbin (21:33.902)

very quickly in terms of getting in front of the customers.

Alp Uguray (21:37.476)

So it's in a way they're reimagining how the processes were done, reimagining how their products were serving their customers to augment AI and then make it better.

Jonathan Corbin (21:46.87)

Yeah. Cause you think of like what had to happen in the past, right? You had to go through these kind of like multiple siloed organizations in order to get a great experience. Let's think about like a B2B example for a second. If you were to go purchase software, let's say it does marketing. and you would like, if you were like, I want to purchase a marketing product, you would go out and you would do your research. So you would go through and you'd look at a bunch of reviews and stuff like that. You start to narrow it down.

you probably start engaging with marketing collateral. Then once you start like, okay, great, this is the company I want to work with, let me engage with them and like, I'm to talk to a sales rep. You kind of like, you know, information and the research and like the marketing collateral that you interact with this capture or, you know, exchange with sales rep, and you're going to have to explain the same reasons of why you're looking for it, that you probably already researched through the marketing collateral.

And then once you talk to the sales rep, you're convinced this is the tool for me. I'm going to buy it. I'm to purchase it. You've given them all your use cases and you have the pricing and all that other fun stuff. And you like sign in the dotted line, then you're handed over to someone who's going to do onboarding for you. And then you have that conversation all over again. And then after you go through onboarding, then you're handed over to like maybe a services team, maybe a CX team. And you have that same conversation all over again. And then when you call them to support, they say, Hey, well, why did you purchase our product?

And you have that same conversation all over again. Right. And you're like, my God, like, how can you guys just get your shit together? Like you have all that information. It's stored someplace. Please combine it into a single entity. So whoever's talking to me has all that context. I think that's what the future is going to look like where we're not going to have these like siloed functions that exist today. You're going to have a singular touch point. It's going to be able to engage with you on a whole bunch of different things that we had to go to different people for before.

Alp Uguray (23:11.61)

Yeah

Jonathan Corbin (23:39.65)

Whether it's like, want to add additional seats so that I can have more people using my product. I need to add additional data. I want you to set up a marketing campaign for me. Or maybe I just need you to do something for me, like set up that campaign and to produce space for me. So I think this is what the future is going to look like. I'm really excited about that.

Alp Uguray (24:03.204)

That sounds very, very exciting. It will make the customer's life easier and also the employees life easier so that they will not have to deal with angry customers, but also deal with like hundreds of different systems that have very siloed data, different interactions. It will completely change.

Jonathan Corbin (24:23.694)

Do you know why the average ticket price for like an enterprise customer interaction is $40 to $60?

Alp Uguray (24:33.712)

No, no way

Jonathan Corbin (24:35.886)

Because they're logging into seven to 10 different systems in order to be able to answer those questions. And if they can't find it in those seven to 10 systems, you know what they're doing? They're going to a Slack channel and they're asking the people who work with them.

And so like that's why it's so expensive. And so when you start to break down those barriers across the silo data and between like finding the information for people and the answers that they need, like your costs become much more efficient in terms of interacting with them, your customer frustration goes away. So yeah, it's pretty cool. I think what we're going to see in the future.

Alp Uguray (25:13.232)

That's crazy the thing about it, that they have to go through 10 different systems to log in. And especially if they have different credentials to remember the passwords. Imagine doing two -factor authentication 10 times. The customer is waiting on the other end. That's really tough. That was really insightful.

Jonathan Corbin (25:22.542)

I circled it into my bank account. Imagine that. That's tough.

Alp Uguray (25:43.642)

That was really insightful. So what's next for Maven AGI? know that customer support is beginning. How do you see it evolve and expand to come what's next?

Jonathan Corbin (25:58.562)

I think when we think about the evolution of customer support, it's that first you start off by answering questions, then it's injecting personalization and being able to take complex actions. And so as we're continuing to grow, there going to be some announcements coming out in the next month or so of some really cool stuff that we're going to be doing to better enable people to ingest more data into their customer interactions and then continue to build out the people that they're working with.

Alp Uguray (26:25.85)

That's exciting. And so I have one other question, actually two other questions. And then this one will be quick. think the first one is what would your advice be to maybe an aspiring entrepreneur for on what to do next or like how to approach a business problem and then find a solution to it.

Jonathan Corbin (26:32.172)

Yes.

Jonathan Corbin (26:52.546)

Yeah, I'm a big fan of taking it down to first principles. Because if you solve the obvious problem, that may not be the one that you actually need to focus on. And so I'm a big fan of take it down to the first principles approach and what's the actual problem that needs to be solved. So that would be the first thing that I would do. And then I think the other thing, it's like you always hear a ton of around like,

you know, hey, you know, go find a large enough tam and do some other things. But I think it's like leaning into areas that you're familiar with, areas where you have an unfair advantage because of your specialized knowledge and skillset is something that I think is not mentioned frequently enough.

Alp Uguray (27:34.798)

And for my last question, I ask every guest what they would like to ask the next guest a question. So what would you like to ask to the next guest? And also, of course, you will have to answer that question too. So it could be anything you're curious about.

Jonathan Corbin (27:51.318)

Yeah

Jonathan Corbin (27:58.006)

Yeah, I think that the thing that I'm most curious about, and I think all of us are envisioning this, it feels a little sci -fi -ish, right? Like, what's the future of gen .ai look like? I think that's a very cerebral question and requires a bit of an imagination. The things that we are doing with generative AI now and the capabilities.

Like we probably read about them in books before, right? I'm sure we all see these like Instagram memes of like 20 years ago in this book, this person said this thing. It was a little bit like that, but it is like, it is kind of, know, science fiction of what we're seeing it do and the capabilities around it. And so I'm really excited about where it's going and what we're going to be able to do in the future.

Alp Uguray (28:48.398)

Yeah, I think there's a big correlation between science fiction and what humans work on.

Jonathan Corbin (28:54.988)

Yeah, Who knows if we'll be able to colonize Mars next?

Alp Uguray (28:59.248)

Maybe we're already trying to send ships and rockets there. That said, this was a fantastic conversation. Thank you very much for your time. I hope I'll see you next in the basketball court again soon. Sunday morning. Good to see you too, Giotto.

Jonathan Corbin (29:12.631)

Enjoyed it!

Jonathan Corbin (29:17.864)

For sure, let's do it. Sunday morning. Great to see you out. Appreciate the time.

Bye.