Harnessing AI for Small Business Success with Constant Contact’s Louis Gutierrez

Hey everyone. Welcome back here to Techstrong tv. Uh, my next guest is Louis Gutierrez.

Louis is the director of AI at Constant Contact, and let's welcome into the show. Louis, great to have you on Tech Drunk tv. How are you?

Great. Yeah, thanks for having me on. Alan.

Excited to be here. My pleasure to have you. Hey, Louis.

I'll be honest, we don't get a lot of director of AI here yet, though I think we'll be getting more and more in the coming months and years. But, um, give people a sense of, you know, what, what's your role as a director of ai? What, what were your qualifications?

Yeah, yeah. Uh, you know, what's your road been like? Yeah, for sure.

So, um, briefly I'll just describe what Constant Contact does. So Constant Contact, um, provides AI enabled digital marketing tools for small and medium sized businesses. Um, so as the director of ai, which I've been at this role for about 10 or 11 months, which is relatively new for Constant Contact, we have engineers that have been here five, seven years.

Um, we have some people that have been here 10 years because we're a company that's been around in the marketing space for, uh, almost 30 years now. Uh, so it's still relatively new, but generally my role breaks up into two different parts. So one of them is ai, integration of AI into the product side, so the customer facing side.

So this, um, manifests itself in content generation tools, brand analyzers, so forth. Uh, and then the other side of it is the productivity side. Like at here, at Constant Contact, we're a medium sized business, uh, and we want to utilize and leverage AI tools to bring productivity gains efficiency to free up people to work on things that they're passionate about, um, while delegating some of the other things over to ai.

So my role, um, generally falls into those two categories. And our strategy is pretty simple here. Uh, we approach AI through a clear-eyed, uh, way.

We want to, we want to make sure that we don't get caught up in the hype of adopting AI tools, um, just for the sake of the new shiny thing, but we want to adopt AI so that it brings value so that it brings and more importantly, measurable value. And so that goes both on the productivity customer facing side and on the internal productivity side as well. Uh, and so we do that by measuring adoption, by measuring ROI, by measuring quality and so forth.

Um, so, so that's a little bit about, uh, what I do here at ai, uh, at constant contact in, in regard as the AI director. So my journey, um, started off, uh, as a software engineer. I did undergrad and, and master's.

And so I worked as a software engineer for a while, and then at a certain point I decided to go back to grad school, and I was actually really interested in cybersecurity. I think you have a background, right? And some experience Yes.

In security. So I went into grad school, um, and to do my PhD, and I was, you know, had an advisor. I was ready to do cybersecurity.

Uh, and then I, you know, as part of, uh, you know, PhD program, first year PhD programs, we had to attend seminars and lectures and conferences. And so I went to one and I saw somebody presents, like on an early on machine learning results, I think they were analyzing, uh, an article from the New York Times and looking for bias and topic modeling. And I remember sitting there and thinking, wow, this is really bad.

This is so bad. There's no way that this is ever gonna work. And so I, I, uh, I, I harassed like, as you do in academia, you harass and give the the speaker a hard time and ask 'em a bunch of technical questions, uh, make 'em sweat up there.

And then afterwards I went and I talked to this person. I was like, wow, this is, you know, so bad. This is, uh, you know, this is never gonna work.

This is never gonna scale. People are never gonna wanna use this. Uh, and then we started chatting and talking.

We went to go eat coffee. And, uh, 15 years later, here I am, right. I've dedicated my, I switched over my major, my concentration from security over to machine learning and AI found a new advisor.

Uh, yeah. And I've been focusing on that since, you know, roughly around 2009, 2010. So it's a, it's pretty a long journey here.

So He converted you, He converted me just one simple conversation, probably 30, 45 minutes, and I was really, was Enough to ignite. Yeah, for sure. It was, it was almost like, maybe this is a little bit of hindsight bias, but it was, he, he convinced me that someday I would be having a conversation like this, right?

That there would be mass adoption, that these tools would be made publicly available, that they'd be easy to adopt, uh, and that, you know, we would be, uh, you know, at the forefront of technology, innovation, productivity, and changing the way that people work. Um, not in so many words and not so specific, but that, that type of, uh, impact on, on society, I think is what he communicated with me. And, and it worked.

And here I am, AI's gain was securities loss, huh? Um, right. But we're here, and, and, and you're right.

I mean, look, this is you. I, I often think this, if we took someone not from the last century, not even like from the nineties, if you just took someone from 10 years ago, 15 years ago and dumped them, you know, in front of a computer with chat GPT today and said, have at it, it, it would be akin to like, you know, primitives praying to a god for rain because they thought that God brought rain. You know?

Um, there are some people who still do think that, but that's a whole nother story. But anyway, you know, it's so, it's almost auto magical, right? Yeah.

In the, in the, and what it does here, I I was having a conversation earlier today, you know, when will we achieve super intelligence? When will we achieve, uh, uh, uh, uh, artificial general intelligence? A GI, and I don't look, as I sit here today, I can't tell you when, or even if we may never, who knows, maybe the, you know, we top it out.

But even if we stopped developing AI where we are and froze it today, there's enough here to revolutionize and disrupt industries for years to come, years to come, that that's how impactful that that is. But, you know, Louis, I've been in in tech a long time, A lot of times, you know, it's the big guys, it's the big boys who get to take advantage of this technology because there's oftentimes a high barrier or of entry. It's expensive.

You need really, you know, boat, you know, crazy good skills. You know, there's something that kind of prohibits the, the every man, the small, the SMB, the mom and bob from harnessing, right? Like I, I remember for instance, when the web first came out, everybody should have a website.

Well, where's a small company going to go get a web? Right? They didn't have you.

You could go hire someone, but that got expensive quick too. Um, AI seems different. It's, it's almost that the big DD well, it spreads democracy, right?

Everyone, everyone could use it. You don't have to be a prompt engineer. You don't have to be a director of ai.

Talk to us what you guys are seeing at Constant Contact. You, I don't know how many customers, but you guys have to have tens of thousands of Mom and Pops and SMB businesses as customers. Yeah.

Yeah, absolutely. So you, you're absolutely right. I think there is something fundamentally different about what AI means, and I think it's even more specifically like our perspective is that there's, there's actually an, an, an added benefit to being a small business.

So if you look at, um, the disruption, especially as far as like job displacement is you look at the big companies, and so I've worked at big companies before, right? Where the, the, the name of the game is like, you have a problem and you throw a bunch of resources and money at it, and it gets solved, right? Um, but it's not necessarily like a long-term solution in some cases, right?

Uh, and I think, um, what these bigger companies are learning with AI is that they can have leaner teams and then enable them with AI tools so that they're, those leaner teams are, they're, they're force, they're force multiplied, right? They have a bigger impact. So they're able to do more with less.

So the interesting thing, uh, and the perspective that we take at Constant Contact is that small companies, small mom and pop shops have, have known this since the beginning. They've worked with small teams and they've made people on their teams, uh, wear multiple hats and do multiple things, right? And so the difference that's happening is here, is now they're enabled with AI tools like Constant Contact that allows them to do more with less.

So in a sense, we, we see AI as bridging that gap or like leveling the playing ground between the big companies and the smaller companies at Constant Contact internally, we use a lot of the same tools that the bigger companies use. We just, we buy licenses like, for example, cur for AI system development. On the product side, it's about content generation, right?

So what we do is that we look at, uh, our goal is that we want our customers to be able, they're not marketing experts. They're, uh, real estate agents, right? They're, they have ice cream shop, they have a bakery, they're experts.

They're passionate about their business, and marketing is something that they have to do at Constant Contact. We enable them with these AI tools so that they can have the, the least amount of efforts that goes into it with the maximum amount of output, right? And we do that by allowing them to generate templates, email campaigns, social media can, um, by having smaller smart input with their brand data, with their campaign history data, with their, with their online presence, we enrich that context, and then we're able to produce templates, email templates, social media posts, SMS, that is tailored to their audience, right?

Where you wouldn't ordinarily get that if you just went to chat GPT and said, you know, I have a, I'm releasing pistachio flavored ice cream this Friday. Write me an email, right? It's gonna be too general purpose.

So we are able to carve out a space for our customers in the marketing space and be able to have very tailored, specific marketing copy generated for them. Excellent. So, again, I've been around sometimes the road to, you know, what they say, the road to perdition is lined with the best of intentions, right?

Right, right. You're a director of ai, you're very comfortable around it. I'm using it.

We're techie people. We we're comfortable with this stuff. I think sometimes the guy owning the ice cream shop or the flower shop or, or what have you, feels a little bit like this stuff is being forced down their throat and they're not comfortable with it.

And so, you know, they use it if they have to, but they prefer not to. Let's say what, but like you said, I don't know if we've seen something as easy as, as wearable as AI is, what, what's been the reception from the constant contact customer base loss? Yeah.

Yeah. So we, we re um, we're coming out with a report pretty soon, um, and we're, we're noticing that essentially 48% of all the SMBs are, are using AI tools. They're integrating them into their life.

Yeah, yeah. Yeah. And, and that's a big amount, I think like when you start to break it down, yeah, it's a big amount, right?

Yeah. And you know, when you start to break it down, emails and social media posts is like 37%, right? Um, so it's, it's a, it's a huge adoption, right?

Um, and I think that the way that we manage it in constant context, so there's, there's a lot of, um, a lot of things you have to watch out for. If you're, if you're out, you're out, out there and you wanna adopt AI tools, you have to think about you, legal, legality, compliance, right? Um, ethics, right?

Responsible ai, and you have to think about, um, uh, a bunch of other things that hallucinations, for example, right? Like, how do I, how do I manage hallucinations? How do I make sure that if I'm generating some sort of marketing content, that it's relevant?

That it's not just something that, that, that the AI made up. Uh, and so like in constant contact, we do that hard work for everyone, right? So from one of the first things that we did before we even started integrating AI into our product is we developed an AI governance team that is a cross-functional effort to security.

It, ai, team legal, we all come together, we decide what the rules are, right? Until the day when government comes in and says, here's the rules, like companies are left these rules, and since we are put the customer first bringing value to the customer, sure. That we're delivering responsible, ethical solutions for them, that's at the forefront of our air governance team.

So that, that's the case hallucinations, right? A big one. How do people p That's the first thing that I hear from people is like, how do I know that it's just not regurgitating up some nonsense to me, right?

Um, and so we do things, we add context, we add different types of mechanisms in there to reduce hallucinations. Like if you think about hallucinations, one of the problems with hallucinations is that a lot of times, one of the reasons it could come up is that the, the model might see training data and then it might see other training data that can flex with it, and then it might just give you some noise, right? That's one of the reasons for it.

Um, so what we do is we make sure that we validate some of our results to minimize. So with any probabilistic model, generative discriminative, whatever the case is, any type of machine learning or AI model, you're going to have error associated with it's inherent, which we do our best to minimize that error, um, through several, uh, you know, uh, several functions within, within us, and then also we monitor it as well. So we have an observability platform that allows us to monitor it for quality.

So we have different quality metrics that we can see how those metrics behave over time. If we start to notice that quality goes below over a certain threshold, then we go in there and investigate what it, um, and then we also, uh, monitor for toxicity, for quality, for all these other things to make sure that we're providing the best experience and minimizing that, that error, that threshold for error for our users. And then, um, you know, also we do, you know, we do all of the legwork from, from legality perspective, like everything on our side goes through legal vetting to make sure that everything is ready to be used on the user's end.

Excellent. So what appears to the user to just be a really easy pick it up and run with it is really pretty heavily monitored, optimized, maintained by you guys. Yeah.

We're almost outta time. I want to bring up that legality thing. Have you guys looked into, because I don't even know what model, are you using a frontier model?

Is this your own LLM is, is the data in there responsibly, harvested, I guess is a good word? You know, what, what's constant contacts take on that? Yeah.

Yeah. So, um, we, we have like a perspective where we put security and ethics first. Like this is, this is really obvious.

Our CTO is a security guy, right? So like, you know, the first question that he has every time is like, how can we make this secure? How secure is this?

And how can we make it more secure, right? So from that aspect, we don't star, we don't store raw input, you know, prompts. We don't store them, right?

Um, so, so that's one aspect. So we never store that data, right? It's just, it's ephemeral, right?

It just, it comes in and then it disappears. Uh, and then also every, everything that we do, like when we work with third party vendors, for example, OpenAI or other third party vendors, we make sure that they go through a rigorous legal and procurement process to make sure that all of the i's are dotted and all the t's are crossed on that end. Um, then when we do train, like, like for example, we have spam detection models that we train internally.

Um, we're experimenting and trying to answer some questions about fine tuning models, so taking open source models and fine tuning those with our, with our own internal data as well. And so all of those then would re with, would live behind our virtual private cloud, right? So it's all within, in the same place where all the user data is.

So it doesn't actually leave like our, our secure space as well. Um, so, so those are some of the precautions we take as well as just we do audits. We have, like we said, we have like these observability platform, so we're always constantly looking at our data and doing manual review of our data, uh, as well as monitoring different metrics that we have to, that can give an indication on whether like the distributions of data might be changing and so forth.

I love it. Yeah. Hey Louis, we're about outta time.

We've gotta wrap up. com, but is there a specific section of the website where you can get all the great AI tools and Yeah, so What I would recommend is, the best thing to do is start a free, free trial. com, start a free trial, and start to play with some of the tools on there, uh, and generate some prompt, like generate some emails, test it for yourself.

That's the best way. One of the things that we want to do is we want this to be very natural and organic for people. So it should be that someone starts a trial and they're very seamlessly go in to be able to generate some emails.

Love it. Hey, thanks for coming up here on Text Drug TV and talking to us. Continued success to you and all the folks at Constant Contact.

You do good work. Thanks so much. Take care, Lewis.

All righty. Thank you. Louis Gutierrez, director of AI for Constant Contact here on Text Drunk tv.

We're gonna take a break. We'll be back in a moment.