Swimming With Sharks: Customer Ops Unplugged - S2 Episode 2: Aaron Jones

In this exciting episode, Kevin Dean sits down with Aaron Jones, the Head of Strategic Partnerships at Maven AGI. The conversation spans Aaron's journey from a computer science major to his current role, and delves into the dynamic landscape of customer operations and AI.

Episode Summary:

Introduction: Kevin Dean introduces the podcast and welcomes Aaron Jones, highlighting their shared background and long-standing professional relationship.

Interview Highlights:

• Aaron’s Background: Aaron shares his career journey, from his early days as a developer to his pivotal roles at major companies like Adobe, Mixpanel, Sprinkler, Discovery Education, HubSpot, and now Maven AGI.
• AI and Customer Ops: Aaron discusses the similarities and differences between the current AI boom and the internet bubble of the 90s. He emphasizes the strategic application of AI and its significant impact on customer operations.
• Challenges in Customer Service: The discussion touches on the three main challenges: increasing customer expectations, limited resources, and the balance between human capital and technology.
• Framework for Success: Aaron outlines a strategic framework for organizations to efficiently manage customer interactions through self-service, AI, and human resources.
• Technology Evaluation: Insight into how businesses can assess and integrate the best technology solutions to enhance customer experience.

Real-life Example:

• Maven AGI and TripAdvisor: Aaron shares a case study where Maven AGI helped TripAdvisor successfully deflect over 90% of inbound customer support cases using generative AI.

Key Takeaways:

1. Core Competency Focus: Emphasize core strengths and leverage best-in-breed solutions for customer support.
2. Data Strategy: Ensure data is interconnected and contextualized for actionable insights.
3. AI Implementation: Proper training and execution of AI models are crucial for success.
4. Embracing Automation: Automation can enhance efficiency and create new job opportunities, rather than just replacing human roles.

Transcript

Kevin Dean (00:01.708)
Hey there and welcome to Swimming with Sharks, a deep dive into customer ops. I'm your host, Kevin Dean, CEO of Manobite, and I'm thrilled to have you join us for this exciting journey through the dynamic world of customer operations. On this podcast, we're going to explore strategies, tools, and innovations that drive exceptional customer experiences. And this week, I am super excited to have my friend, Aaron Jones,

Kevin Dean (00:29.964)
Head of Strategic Partnerships at Maven AGI joining us for a really fun conversation. Aaron, great to have you on the show today.

Aaron (00:39.498)
Kevin, it is my pleasure to be here joining you and looking forward to a great conversation.

Kevin Dean (00:45.932)
For sure, for sure. Now, Aaron, you and me, we go back a little bit from your time at HubSpot, but the audience may not know a ton about you. So can you give them a little bit about yourself and your journey?

Aaron (00:58.733)
Yeah, absolutely happy to. So I was a computer science major in college. I started my career as a developer and did all kinds of geeky things. I like to affectionately refer to myself as a recovering geek. But things have advanced so far and I've seen so many incredible engineers now that, you know, like my days of writing code are like so far in the past. But it was like...

Kevin Dean (01:09.1)
Thank you.

Aaron (01:26.924)
good way to get started. And I took that experience and leveraged that to move into data science. And me and another gentleman found it, which ultimately became the data science program for the Discovery Channel. And ran that for a while. And from there joined Adobe, which is what got me from sort of the in -house side.

to the vendor side of the house. And that was my first foray into the SaaS world. And then from Adobe, there was Mixpanel, Sprinkler, Discovery Education, HubSpot, a few other spots along the way. And that all led me over to Maven AGI, which I could not be more excited to be in this exciting space.

Kevin Dean (02:20.108)
yeah, I am super excited to learn more about that. I did not actually realize that your degree was in computer information science. That's the same as me. That's how I started as well. I didn't realize that. Yeah, so data geek from way back. But yeah, like you said, like I've got other people who do that for us now. Don't put me in front of a computer. But things are changing so fast. I mean, wow. I mean, I think about

Aaron (02:29.642)
Yeah

Yeah

Aaron (02:36.49)
Yeah.

Aaron (02:42.475)
Right.

Kevin Dean (02:49.932)
about the time probably when you and I were going back doing all this stuff and where we are today. Man, what a difference, right?

Aaron (02:57.418)
It is crazy. I remember when JavaScript was invented and I actually was pretty good at JavaScript once upon a time. But things have just moved so fast and the world is just so different from the time when I was building systems and writing code.

Kevin Dean (03:15.564)
Yeah, it's completely different. Now you and I probably remember the internet bubble. We remember the internet bubble. Let me ask you this. Do you think what we're seeing right now with AI is similar to the internet bubble? Yes or no? Why or why not?

Aaron (03:23.914)
yeah.

Thank you.

Aaron (03:40.745)
Now that's a great question and I would love to give you a yes or a no, but I'm gonna I'm gonna nuance it a little bit I Think there are similarities Between the two but I don't think that they're necessarily similar but there are similarities and what I mean by that is back in the 90s The internet was like this new exciting thing and if you were on the internet that meant that you

Kevin Dean (03:48.268)
Okay, alright.

Kevin Dean (03:56.972)
Okay, okay.

Aaron (04:10.409)
were a successful business. But during that time, as we both lived through, there were many businesses that didn't have a business plan. Like they didn't have a product. They weren't really doing anything. They were just, you know, an entity or a presence on the internet. And that's not a business, right? And a lot of those companies washed out when the bubble burst. Today, there are lots of people in the AI space.

And there are many practical applications. I think the similarities is that I think that, from my perspective, is there are many companies that are in AI, but that's not their core competency and that's not with their businesses. They understand that it is here and it's not going anywhere. It's going to continue to evolve and get better.

And I think that that's where the similarity is. I think there are a lot of people throwing resources at AI, but not necessarily strategically. And I mean that with all due respect, you know, but I think that it's like, there are a lot of companies that are getting into it, but that's not what they do. And ultimately focusing on core competencies is going to drive success for everyone.

Kevin Dean (05:18.38)
Mm -hmm. Of course. Of course.

Kevin Dean (05:26.956)
Right.

Kevin Dean (05:35.564)
Yeah, I think you're spot on with that and I appreciate those nuances. You know, one of the things that I'm seeing a lot of is that anybody who knows how to put in a prompt into chat GPT is now an AI consultant.

Aaron (05:49.319)
Right. Exactly. Yeah.

Kevin Dean (05:53.132)
But, but, but, but, you know, there are real challenges that are out there, especially in the world of customer operations that, Gen AI and AI can solve. what are some of the big challenges that you see out there today that are facing companies as they're trying to really deliver great customer experiences?

Aaron (06:16.994)
So I think that there are sort of three prongs to the challenges that are out there. Customer expectations are continually increasing, right? Tolerance for inefficiency or patience for not getting to that thing that you need as a customer. Like that tolerance is decreasing as expectations are increasing. And then the third piece is they're just limited resources.

like humans can only scale so far and human capital can only do so much. And the challenge is how do you balance human capital, human resources and technology the right way so that we can meet those increasing expectations quickly? Because again, that tolerance for inefficiency or for waiting to get to an answer or to be being handed off multiple times before getting to like a final

Solution or resolution? It's it's tough, but I think that it's like try threading three needles at once

Kevin Dean (07:24.46)
Hmm. Yeah, that that is tricky. and I don't know, do you think there's a fourth needle that really is around understanding of what's available to companies and what what resources are and just having knowledge? Is that another needle that you think might be needing to be threaded as well?

Aaron (07:45.696)
Yeah, I think that's a great point. There are so many ways to address customer challenges. And so what's best for your business and what's best for your customers? And like sort of what's the right recipe to deliver that great customer experience that's efficient, that's fast, and it gets customers the solutions that they need.

not just responses, but the actual solutions. And also that it's a good experience that they're not being handed off multiple times, that they're not having to repeat themselves or sort of once they get to that next rep or that next person, they have to start at the beginning and explain everything all over again. These are the challenges, but what's right for your business and what's right for your customers?

may be a little bit different depending on your industry.

Kevin Dean (08:47.372)
You know, that's a great way to think about it. And you've led large customer organization, customer service organizations in the past. Do you think that there's a framework for organizations to kind of look at how to start strategically thinking about what they need to do to meet those needs?

Aaron (09:07.97)
Absolutely, and there is a framework that I think is I'm not going to say it's universally applicable, but it applies to most most businesses and most most install bases. It's like there are Oftentimes the 80 -20 rule where 80 % of the questions come from 20 % of the scenarios, right? and then understanding of those things

When are humans absolutely necessary? And where can automation or like now everyone is looking at rightfully so, when can AI play a part or when can customers self -service? So it's like really having a clear understanding as to what those breakdowns are and how to segment out those things and then how to like really efficiently.

route customers to the appropriate channel. So if it's something that they can easily self -service, let's make sure that it's easy for them to self -service. If it's something that there's a UI where AI can help them out, whether it's directly through an AI -powered search or a co -pilot that's assisting a person, that's easy to get to.

Kevin Dean (10:30.796)
Mm -hmm.

Aaron (10:34.143)
And then also if it's something that no, this is just so complex or nuanced that we need to get a human resource involved that getting to that right resource is something that's easy for the customer.

Kevin Dean (10:47.724)
Well, that's a nice framework to think about. So included in that framework, you mentioned it, there's a lot of technology plays. How do customers go about evaluating what's the best technology fit for them and their needs? How would they go about doing that?

Aaron (11:08.607)
Great question. And that's, this is where I always come back to like core competency. There are customers that's, there are offerings and opportunities for just about everything. And there are some shops that sort of start to expand and reach into many areas. And...

You know, there are jacks of all trades and masters of few. And then there are some folks that are very highly specialized. And I think that having an opinion, having a strong opinion on how you want to go about solving for the customer experience is important. And for me,

Kevin Dean (11:57.356)
Okay.

Aaron (12:00.062)
And a philosophy that I hold is I prefer to go with best in breed for whatever it is that I need a solution for. And so for customer support, I would think about who do I believe is best in breed for solving for customer support. And if it's something else, what is the best solution for it?

And if there are clear leaders in the space or if there are leaders that are defining a space, let's look into that and let's make sure that we are sort of future -proofing the way we're taking approach for solving for the customer experience.

Kevin Dean (12:47.916)
Yeah, that's great. I like that a lot. That's a good approach. I think that best in breed is really a good way to go, especially, you know, it's funny, you can buy one really nice pair of shoes, or you can buy 10 pair of cheap shoes.

Aaron (13:10.108)
You know Kevin, it's funny that that's the analogy that you use because God bless her soul. My grandmother used to always tell me that she's like Aaron never ever ever buy cheap shoes because they will wear out they won't be comfortable and you're just gonna get a new pair spend the money get high quality and It will last you much much longer and it'll actually be more comfortable to wear. It's it's literally

Kevin Dean (13:21.9)
fuck it.

Aaron (13:40.123)
like the perfect analogy.

Kevin Dean (13:42.412)
Yeah, yeah, it's so true. It's so true. So let me ask you this. Let's talk a little bit about how GNI is solving for customer problems. Can you give me some real life examples of how GNI is really solving for customer problems? And yeah, I don't want to hear about, you know, generating a pretty image or, you know, writing up writing up.

Aaron (14:09.242)
Yeah.

Kevin Dean (14:11.468)
my email, give me some real examples of how this is being done today.

Aaron (14:16.666)
You know, this is why I've actually joined Maven AGI is because of what they're doing and how they're leveraging generative AI specifically for customer support. I'll give you actually a real life example. There's a customer TripAdvisor that's using Maven AGI.

And TripAdvise has been around for a while. They've got a massive network. They've been helping people figure out their travel plans, make recommendations, book hotels, restaurants, the whole gamut for years, decades probably. Since leveraging Maven AGI and training the model, I think I've got this right, but I believe that they're deflecting

over 90 % of their inbound, like successful, successfully deflecting over 90 % of their inbound customer support cases. So that is, that's pretty powerful. Like that's really, really significant. I am just like blown away by when executed and implemented correctly.

Kevin Dean (15:13.068)
Wow

Kevin Dean (15:20.843)
Wow.

Aaron (15:38.296)
how incredibly powerful generative AI can be. But like those are two pretty big caveats when, you know, implemented correctly and well executed. Like that's like, you can't take those things for granted. Like you've got to get it right. It's as good as it's trained. And it's not magic. And I think that that's something like is awesome and as much potential as AI, particularly gen AI has.

Kevin Dean (15:55.532)
Yes.

Aaron (16:07.959)
It's not magic. There is a, it's as much art as it is science and making sure that it's set up correctly to deliver those types of results. And it's, you know, I'll do a shameless plug, but that's one of the things that Maven AGI is really sort of focused on is quality. You know, zero hallucinations and making sure that the customers are empowered.

Kevin Dean (16:32.172)
Yeah.

Aaron (16:37.559)
to see if there is an area that needs to be improved upon that is presented to the customer so that it can easily be corrected so that in the future it won't be a question anymore.

Kevin Dean (16:51.244)
I love that. So I'm going to ask you this first question so I can ask you the second question. So there are a lot of people who don't understand what training and AI model means. Can you give me a high level overview of what it means to train a model?

Aaron (16:55.318)
All right.

Aaron (17:08.246)
Absolutely. To train a model means giving it the appropriate and accurate information to handle prompts or inputs. I'll give an example. ChatGBT was the first sort of like publicly available commercial application of generative AI. ChatGBT is great.

It really is. However, it basically learned the entire internet just directionally, right? Like it just sort of sucked in everything. Some of the things that it was trained on, that it learned on, were not necessarily accurate. Therefore, exactly, or it's out of context. And when that occurs,

Kevin Dean (17:58.092)
Or out of context.

Aaron (18:06.709)
There's something that's called hallucinations that can be output and it sounds right, but the information that the language model learned was either taken out of context or it was inaccurate and therefore the response that's given is not solid. And so it's important that when you train a language model...

that it's got accurate current contextualized information. So that way when it's prompted, it will give the correct answers. And there will be times. I mean, that's part of the process because training a model isn't one and done. You will train it today, but then your product or services will evolve over time. So it has to be updated and retrained and it's going to be a new

business process is the way I think about it. It's part of the support and maintenance so that you can maintain those sort of efficiencies that you gain by having this thing available that can handle incoming requests. But it's only going to be as effective as it's trained. So the old, and I'm sure you remember from back in our coding days, there's garbage in, garbage out.

Right? If you train it on good information, you will get great responses. If the information becomes either stale or outdated, then the responses will be based on that information that is now stale and outdated.

Kevin Dean (19:52.524)
Yeah, that's powerful. So, you know, a while ago we used to hear this a lot that data is king. Has data become king again?

Aaron (20:05.301)
Data is king in a different kind of way. I think it's like a more powerful king because, you know, back in the days when I founded the data science program at Discovery, well, co -founded it, we had to sort of take data, process it, create insights, and then push those insights out to...

Kevin Dean (20:12.396)
Okay, okay.

Aaron (20:33.909)
business owners to general managers and and and VPs of the different business lines and that worked for a couple decades but as things have evolved you know people want to self -service people want to be able to get to the insights that they want and once they get that insights they're going to want to dig further and get additional insights and that's one of the things that makes AI and

generative AI is so powerful is that you can continually sort of prompt and enhance the prompts and ask additional prompts and get back feedback. So the data that these models are trained on that provide those responses has never been more powerful or important.

Kevin Dean (21:26.444)
Yeah. So, you know, we work with several large organizations, you know, billion dollars in revenue plus. And a lot of these organizations still have disconnected data systems. How important is it for organizations to really start thinking about, one, ensuring they've got all their data connected, and two, throwing this out there, trying to get onto

Make sure everything's in the cloud.

Aaron (21:58.933)
I love this question. So, folks that have worked with me have heard me say many, many times, data without context is trivia.

And it is. You need context. So often data exists in silos. Those silos, when joined together, provide context for each of those data points or for all of those insights. And contextualized data is what's actionable. Data that's just on its own, just sort of siloed and without context, again, to me, it's just trivia.

Kevin Dean (22:16.556)
Mm -hmm.

Kevin Dean (22:31.788)
Mmm.

Kevin Dean (22:39.692)
Mmm.

Aaron (22:39.797)
It's not really meaningful and it's absolutely not actionable. So interconnecting as many data sources as possible, making that available sort of in some type of cloud infrastructure so that it's available to any of the applications that may need it or to help train these language models. Like that's where things are today.

And if you are not there today, it's probably a good idea to get there as quickly as possible so you don't fall behind.

Kevin Dean (23:15.692)
Yeah, I think that that's so important. That's super helpful. Wow. This has been so helpful. So let me ask you this. If you were to share with our listeners one thing that you think is critical for them to go and do today, what would that one thing be for them to go and do?

Aaron (23:43.981)
So, you know, shout out to Manobite, HubSpot, Elite Partner, amazing work that you do with your install base. If I'm speaking to the customers with whom you're working, I would say have a very clear data strategy and understand where all your data sits today. And just have like an honest conversation about whether or not it's really...

sort of joined in a way that it can provide actionable insights. That's the first thing. And then really think about how to leverage your human capital, most effectively in efficiency for what they do best. And then really just have like the, you know, an honest conversation about like what can be automated.

you know, the, the elephant in the room that we actually haven't talked about yet is like, there's like a little bit of fear and hesitation around AI. It's like, is it going to take over the world? Is it coming from my job? Et cetera. and you know, once upon a time there were folks that, were in the horseshoe manufacturing business and they made saddles. Right.

Kevin Dean (24:42.764)
you

Aaron (25:05.42)
and there were leaders in that space. And then automobiles came along and it created an entirely new industry. So as you're thinking about how to deploy your capital, your human capital, and where it's going to be most effective and efficient, that's great. Don't be afraid of automation. It will shift the way things work.

But in doing so, it's going to create additional opportunities. We touched on it earlier. These models, they're not magic. They have to be cared for. They have to be tuned. The models have to be trained and then retrained and then kept current. These are functions that have not existed prior. So there are new job opportunities. There are things that will make when you embrace AI,

Kevin Dean (25:51.052)
Mm -hmm.

Aaron (26:00.428)
you will become a much more valuable resource to your business and to your customers because it can do things at a scale that people just can't do.

Kevin Dean (26:15.756)
That's great advice, great advice, and I appreciate this. Man, this has been nice geeking out with you a little bit today. Before I let you go, I just wanna chat with you about two other things, kind of off the geek topic. So when you're not working hard, what types of things do you like to do?

Aaron (26:33.548)
great question. I love to eat, Kevin. I'm not even gonna, I'm not kidding. I love food. All cuisines, I skew a little bit more towards seafood, but I like everything. Like I like a great steak as much as anyone, but seafood is something that I eat a lot of. I'm really, really into

Kevin Dean (26:39.468)
cool.

Kevin Dean (26:48.332)
Mmm.

Aaron (27:03.244)
Fitness, I love riding and cycling. I like doing strength training. I do generally about one half marathon every year, preferably in a place that I've never visited before. I've got a friend that we sort of schedule these things out a year in advance and we'll just travel someplace and do a half marathon. It's a way to try to keep fit as father time.

you know, starts to catch up with all of us. But yeah, that's how I like to spend the time when I'm not helping customers or working.

Kevin Dean (27:33.324)
Yep, yep. Yep, yep.

Kevin Dean (27:46.284)
that's good. That's good. Now, a lot of people know, you know that I love music. Hit me with one song that you think I need to go and listen to today.

Aaron (27:58.956)
that's a great one. So I'm going to, I'm going to do a throwback because I love music also, but I'm, I'm a little more old school. And, just recently I was listening to, the old classic reasons from earth, wind and fire. And like, that's just a happy song. It just makes me feel good.

Kevin Dean (28:16.012)
yeah. Yeah. Yeah, that's a good one. That is a good one. Don't ask me to sing it. I'm gonna go listen to it after the show. But that's a great one. Aaron, it has been awesome talking to you. Great insights. Appreciate it. And I'm sure that we'll connect again soon.

Aaron (28:24.94)
No.

Aaron (28:38.476)
Kevin, it's always a pleasure. I'm looking forward to talking to you the next time. Take care, you too. Bye.

Kevin Dean (28:42.732)
All right, take care now.